Image filtering apparatus and method, and video encoding/decoding apparatus and method using the same

(56) References cited:

WO-A2-2010/039822 • WIEN M: Variable block-size transforms for H.264/AVC, IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEÓ TECHNOLOGY, IEEE SERVICE CENTER, PISCATAWAY, NJ, US, vol. 13, no. 7,1 July 2003 (2003-07-01), pages 604-613, XP011099253, ISSN: 1051-8215, DÓI:

10.1109/TCSVT.2003.815380 • 5.2 Signal Enhancement In: OHM J-R: Multimedia Communication Technology, 1 January 2004 (2004-01-01), Springer, XP002492449, ISBN: 978-3-540-01249-8 pages 180-183, * page 181, paragraph 3 - page 182, paragraph 1 *

Note: Within nine months ofthe publication ofthe mention ofthe grant ofthe European patent in the European Patent Bulletin, any person may give notice to the European Patent Office of opposition to that patent, in accordance with the Implementing Regulations. Notice of opposition shall nőt be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).

Printed by Jouve, 75001 PARIS (FR)

ΕΡ 3 119 091 Β1

Description [Technical Field] [0001] The present disclosure relates to an image filtering apparátus and method and a videó encoding/decoding apparátus and method using the same. More particularly, the present disclosure relates to an image filtering apparátus and method that can be applied to an apparátus for searching a suitable subblock type for when an Μ χ N (M and N: natural numbers) macroblock is used in videó encoding/decoding and performing compression and reconstruction through block transform/quantization suitable for the searched type, and can improve videó encoding/decoding efficiency by recognizing a transform size determined through a bitstream or various information and applying filtering to a transform boundary with the number of pixels to be filtered and the method of filtering varied according to the transform size, and relates to a videó encoding/decoding apparátus and method using the same.

[Background] [0002] The statements in this seetíon merely provide background information related to the present disclosure and may nőt constitute prior art.

[0003] Moving Picture Experts Group (MPEG) and Videó Coding Experts Group (VCEG) together stepped ahead of the existing MPEG-4 Part 2 and H.263 standard methods to develop a better and more excellent videó compression technology. The new standard is called H.264/AVC (Advanced Videó Coding) and was released simultaneously as MPEG-4 Part 10 AVC and ITU-T Recommendation H.264. Transforms using the H.264/AVC standard are known from Wien M: Variable block-size transforms for H.264/AVC, IEEE transactions on circuits and systems for videó technology, 2003. Moreover, document WO 2010/039822 A2 refers to videó coding using transforms bigger than 4x4 and 8x8. [0004] In the H.264/AVC standard, a residual signal is generated by performing an intra/inter prediction process in units of a macroblock having various types of subblocks, and encoding is performed after further reducing the number of bits by performing a transform/quantization process on the residual signal generated. In a conventional encoding method based on a macroblock, an encoder divides an input image in a 16x16 macroblock, generates a residual block by predicting each macroblock by a size of a sub-macroblock available according to an inter/intra mode, generates a frequency coefficient by applying integer transform designed based on 4x4 or 8x8 diserete cosine transform (DCT), to the residual block generated, and quantizes the frequency coefficient according to a predetermined quantization paraméter (QP). In addition, a blocking effect caused by the transform/quantization process is reduced through loop filtering. [0005] Loop filtering in H.264/AVC (deblocking filtering) is performed in units of a macroblock such as a 16x16 block, an 8x8 block, and a 4x4 block. The main purpose of loop filtering is to remove a blocking effect, and such a blocking effect is generated nőt in units of a macroblock bút in unit of a transform. Since H.264/AVC performs loop filtering along boundaries of a 16x16 macroblock and 4x4 and 8x8 blocks, it is unsuitable for an MxN block size (M may be equal to N) and P x Q transform (P may be equal to Q) and has a problem of applying a filtering coefficient and depth (the number of pixels to be filtered) that is unsuitable for P x Q transform, to a filtering boundary. In addition, it has a problem of failing to variably apply a filtering strength and the number of pixels, as a block size and a transform size increase. Therefore, when various block sizes and transforms are used in high-quality videó encoding/decoding technology to be developed in the future, a blocking effect cannot be effectively removed by loop filtering, thus leading to degradation of subjective/objective performance.

[Disclosure] [Technical Problem] [0006] Therefore, tosolvethe above-mentioned problems, the present disclosure seeksto improve encoding/decoding performance by using various types of reievant square or rectangular transforms in performing encoding in units of a macroblock suitable for a videó, performing loop filtering on each transform boundary by recognizing information on various transform types applied to each block by using a bitstream or various types of information, and applying various types of filtering by determining the effective number of filtering pixels and an effective filtering method according to a block size and a transform size.

[Summary] [0007] An embodiment ofthe present disclosure provides a videó encoding/decoding apparátus including: a videó encoding unit for generating a predicted block by predicting a current block, generating a residual block by subtracting the predicted block from the current block, transforming/quantizing the residual block according to a block type ofthe

EP 3 119 091 Β1 current block, generating encoded videó data by encoding a transformed/quantized residual block, decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual block, generating a reconstructed block by adding the predicted block to a decoded residual block, and filtering a boundary region between the reconstructed block and an adjacent block according to a transform type; and a videó decoding unit for outputting a transformed/quantized residual block by decoding encoded data, decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual block according to a transform/quantization type, generating a predicted block by predicting a current block, reconstructing the current block by adding a decoded residual block and the predicted block, and filtering a boundary region between a reconstructed current block and an adjacent block according to the transform/quantization type.

[0008] Another embodiment ofthe present disclosure provides a videó encoding apparátus including: a predicting unit for generating a predicted block by predicting a current block; a subtracting unit for generating a residual block by subtracting the predicted block from the current block; a transforming/quantizing unit for transforming/quantizing the residual block according to a block type of the current block; an encoding unit for generating encoded videó data by encoding a transformed/quantized residual block; an inverse quantizing/transforming unit for decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual block; an adding unitforgenerating a reconstructed block by adding the predicted block to a decoded residual block; and a filtering unit for filtering a boundary region between the reconstructed block and an adjacent block according to a transform type.

[0009] The filtering unit may determine whether a boundary between the reconstructed block and the adjacent block is a transform boundary.

[0010] If the boundary is a transform boundary, the filtering unit may filter the boundary region.

[0011] If the boundary does nőt correspond to an edge of a videó, the filtering may be performed.

[0012] The filtering may determine a boundary strength according to the transform type.

[0013] The filtering unit may determine a boundary pixel involved in the filtering according to a transform size.

[0014] Rate-distortion (RD) costs for a plurality of transform types of the residual block may be calculated, and a transform type with the least RD cost calculated may be seleeted as a transform/quantization type.

[0015] A size of one side of a block transformed/quantized may be equal to or larger than 16.

[0016] The transforming/quantizing unit may generate information on a transform/quantization type.

[0017] The filtering unit may perform filtering in the same order as in a videó decoding apparátus.

[0018] Yet another embodiment ofthe present disclosure provides a videó decoding apparátus including: a decoder for outputting a transformed/quantized residual block by decoding encoded data; an inverse quantizing/transforming unit for decoding the residual block by inverse-quantizing/transforming the transformed/quantized residual block according to a transform/quantization type; a predictor for generating a predicted block by predicting a current block; an adder for reconstructing the current block by adding a decoded residual block and the predicted block; and a filtering unit for filtering a boundary region between a reconstructed current block and an adjacent block according to the transform/quantization type.

[0019] The filtering unit may determine whether a boundary between the reconstructed block and the adjacent block is a transform boundary.

[0020] If the boundary is a transform boundary, the filtering unit may filter the boundary region.

[0021] If the boundary does nőt correspond to an edge of a videó, the filtering may be performed.

[0022] The filtering unit may determine a boundary strength according to the transform type.

[0023] The filtering may determine a boundary pixel involved in the filtering according to a transform size.

[0024] The transform/quantization type may be seleeted according to a transform type included in the encoded data among a plurality of transform types.

[0025] A size of one side of a block transformed/quantized may be equal to or larger than 16.

[0026] The inverse transforming/quantizing unit may generate information on the transform/quantization type.

[0027] The filtering unit may perform filtering in the same order as in a videó encoding apparátus.

[0028] Yet another embodiment ofthe present disclosure provides an image filtering apparátus including: a boundary identifying unit for identifying a boundary between two transform blocks included in an image; a pixel/strength selecting unit for selecting one or more pixels to be filtered, according to a size of at least one transform block among the two transform blocks; and a filtering unit for filtering at least one pixel included in a region adjacent to the boundary.

[0029] The size ofthe at least one transform block may be a length ofthe at least one transform block in a perpendicular direction with respect to the boundary.

[0030] The size ofthe at least one transform block may be proportional to the number of pixels to be filtered.

[0031] The size ofthe at least one transform block may be proportional to the filtering strength.

[0032] Yet another embodiment of the present disclosure provides a videó encoding/decoding method including: generating a predicted block by predicting a current block, generating a residual block by subtracting the predicted block from the current block, determining a transform/quantization type seleeted according to a block type ofthe current block, transforming/quantizing the residual block according to the transform/quantization type determined, generating encoded

EP 3 119 091 Β1 videó data by encoding a transformed/quantized residual block, reconstructing a residual block by inverse-quantizing/transforming the transformed/quantized residual block, generating a reconstructed block by adding the predicted block to a reconstructed residual block, and filtering a boundary region between the reconstructed block and an adjacent block aecording to the transform/quantization type; and outputting a transformed/quantized residual block by decoding encoded data, decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual block aecording to a transform/quantization type, generating a predicted block by predicting a current block, reconstructing the current block by adding a decoded residual block and the predicted block, and filtering a reconstructed current block aecording to the transform/quantization type.

[0033] Yet another embodiment ofthe present disclosure provides a videó encoding method including: generating a predicted block by predicting a current block; generating a residual block by subtracting the predicted block from the current block; determining a transform/quantization type selected aecording to a block type of the current block, and transforming/quantizing the residual block aecording to the transform/quantization type determined; generating encoded videó data by encoding a transformed/quantized residual block; reconstructing a residual block by inverse-quantizing/transforming the transformed/quantized residual block; generating a reconstructed block by adding the predicted block to a reconstructed residual block; and filtering a boundary region between the reconstructed block and an adjacent block aecording to the transform/quantization type.

[0034] Yet another embodiment ofthe present disclosure provides a videó decoding method including: outputting a transformed/quantized residual block by decoding encoded data; decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual block aecording to a transform/quantization type; generating a predicted block by predicting a current block; reconstructing the current block by adding a decoded residual block and the predicted block; and filtering a reconstructed current block aecording to the transform/quantization type.

[0035] Yet another embodiment of the present disclosure provides an image filtering method including: identifying a boundary between two transform blocks included in an image; and filtering at Ieast one pixel included in a region adjacent to the boundary, wherein the filtering step includes selecting one or more pixels to be filtered, aecording to a size of at Ieast one transform block among the two transform blocks.

[Advantageous Effects] [0036] Aecording to the embodiments of the present disclosure as described above, in an apparátus performing transform/quantization ofa PxQ size and an MxN macroblock avaiiable in a videó encoding/decoding apparátus, filtering is performed on all boundaries where a blocking effect is generated at a macroblock boundary and a boundary between transforms, thereby improving the subjective/objective videó quality and performance ofthe encoding/decoding apparátus as compared to the conventional method.

[Description of Drawings] [0037]

FIG. 1 is a block diagram illustrating a schematie configuratíon of a videó encoding apparátus aecording to an embodiment of the present disclosure;

FIGs. 2 to 4 are diagrams illustrating intra prédiction modes aecording to macroblock types used in typical videó encoding;

FIG. 5 is a diagram illustrating an inter prédiction mode aecording to a macroblock type used in typical videó encoding;

FIG. 6 is a diagram illustrating a method of determining an MxN macroblock type and transform type aecording to an embodiment of the present disclosure;

FIG. 7 is a diagram illustrating a process of identifying a filtering boundary and strength aecording to an embodiment ofthe present disclosure;

FIG. 8 is a diagram illustrating an example ofa filtering method for large-block transform aecording to an embodiment ofthe present disclosure;

FIG. 9 is a block diagram illustrating a schematie configuratíon of a videó decoding apparátus aecording to an embodiment of the present disclosure;

ΕΡ 3 119 091 Β1

FIG. 10 is a flow diagram illustrating a videó encoding method according to an ernbodiment ofthe present disclosure;

FIG. 11 is a flow diagram illustrating a videó decoding method according to an ernbodiment ofthe present disclosure;

FIG. 12 is a block diagram illustrating an image filtering apparátus according to an ernbodiment of the present disclosure; and

FIG. 13 is a flow diagram illustrating an image filtering method according to an ernbodiment ofthe present disclosure.

[Detaiied Description] [0038] Hereinafter, embodiments ofthe present disclosure will be described in detail with reference to the accompanying drawings. In the following description, like reference numerals designate like elements although they are shown in different drawings. Further, in the following description ofthe present embodiments, a detaiied description of known functions and configurations incorporated herein will be omitted forthe purpose of clarity.

[0039] Additionally, in describing the components ofthe present disclosure, there may be terms used like first, second, A, B, (a), and (b). These are solely for the purpose of differentiating one component from the other bút nőt to imply or suggest the substances, order or sequence ofthe components. If a component were described as 'connected’, ’coupled’, or 'linked’ to another component, they may mean the components are nőt only directly 'connected’, ’coupled’, or 'linked' bút alsó are indirectly 'connected', ’coupled’, or 'linked’ via a third component.

[0040] FIG. 1 is a block diagram illustrating a schematic configuration ofa videó encoding apparátus according to an ernbodiment.

[0041] For encoding videós, a videó encoding apparátus 100 may include a predictor 110, a subtracter 120, a transformer/quantizer 130, a scanner 140, an encoder 150, an inverse quantizer/transformer 160, an adder 170, and a filter 180. The videó encoding apparátus 100 may be a personal computer or PC, TV, notebook or laptop computer, personal digital assistantor PDA, portable multimédia playerorPMP, Playstation Portableor PSP, mobile communication terminál or digital TV, and may represent a variety of apparatuses equipped with, for example, a communication deviee such as a modem for carrying out communication between various devices orwired/wireless communication networks, a memory for storing data and various programs for coding images, and a microprocessor for executing the programs to effect operations and Controls.

[0042] An input videó to be encoded may be inputted in units of a block, and the block may be a macroblock. In an ernbodiment ofthe present disclosure, a type ofa macroblock may be Μ x N. Herein, M and N may be a natural number havíng a value of 2n (n: an integer equal to or greater than 1). Specifically, M and N may be greater than 16 and may be different integers or the same integer. In addition, different types of blocks may be used for respective frames to be encoded, and information thereon, that is, information on a block type may be encoded in each frame so that a videó decoding apparátus can determine a block type of a frame to be decoded when decoding encoded data. A type of a block to be used may be determined by encoding a current frame intő various types of blocks and selecting a block providing the optimál efficiency, or by analyzing the characteristic of a frame and selecting a block type according to the characteristic analyzed. For example, if a videó of a frame has a high horizontal correlation, a horizontally-long block may be selected; and if a videó of a frame has a high vertical correlation, a vertically-long block may be selected. [0043] To this end, the videó encoding apparátus 100 may further include a block type determiner (nőt illustrated) for determining a block type, encoding information on the block type, and including the result in encoded data.

[0044] The predictor 110 generates a predicted block by predicting a block to be currently encoded in an input videó (hereinafter referred to as a current block). Specifically, the predictor 110 generates a predicted block havíng a predicted pixel value as a pixel value of each pixel, by predicting a current block in an input videó by intra predictionor inter prediction. [0045] In order to optimize the predicted pixel value, the block may be divided intő smaller blocks prior to prediction, if necessary. That is, a predicted block may be generated in units of subblocks intő which a block is divided. Herein, as described above, the block may be an Μ x N square or rectangular block, and the subblock may be a P x Q block (P and Q may be different from or equal to each other) havíng a vertical/horizontal size of 2n within the rangé of a size of a block (or macroblock).

[0046] The subtractor 120 generates a residual block by subtracting a predicted block from a current block. Specifically, the subtractor 120 generates a residual block with a residual signai by calculating a difference value between an original pixel value of each pixel of a current block and a predicted pixel value of each pixel of a predicted block.

[0047] The transformer/quantizer 130 determines a transform/quantization type according to a block type ofa current block, and transforms/quantizes a residual block according to the transform/quantization type determined.

[0048] Herein, the sizes of a current block, a predicted block and a residual block may be different from the size of a transform block to be transformed/quantized. That is, the size ofa transform block to be transformed/quantized may be selected within the rangé of the size of a residual block. Herein, the transform block refers to a block corresponding to

EP 3 119 091 Β1 a transform unit, and includes transform coefficients or pixel values. For exampíe, the transform block refers to a P x Q transform coefficient block encoded by PxQ transform, or a P x Q pixel block decoded by P x Q inverse transform. [0049] The videó encoding apparátus 100 according to an embodiment ofthe present disclosure may transform a residual block by a pluraíity of available transforms such as 4x4, 8x4, 4x8, 8x8, 16x8, 8x16, and 16x16, and then select the transform having the highest encoding efficiency among them.

[0050] For exampíe, if intra prediction or inter prediction is performed in units of a 16x16 block, all of a current block, a predicted block and a residual block have a size of 16x16. When receiving a 16x16 residual block, the transformer/quantizer 130 may divide the 16x16 residual block intő two 16x8 subblocks and perform 16x8 transform to output two 16x8 transform coefficient blocks.

[0051] The transformer/quantizer 130 transforms a residual signal of a residual block intő a frequency domain to generate a residual block with a transform coefficient, and quantizes the residual block to generate a transformed/quantized residual block with a quantized transform coefficient.

[0052] When the transformer/quantizer 130 transforms/quantizes a residual block, since a transform process is included in a quantization process, the transform is nőt completed until the quantization is completed. Herein, a technique to transform a spatial-domain videó signal intő afrequency-domain signal, such as Hadamard Transform or Discrete Cosine Transform Based Integer Transform (hereinafter simpiy referred to as integer transform), may be used as the transform method, and various quantization techniques such as Dead Zone Uniform Threshold Quantization (DZUTQ) and Quantization Weighted mátrix may be used as the quantization method.

[0053] Thescanner 140 generates a quantized transform coefficient string by scanning quantized transform coefficients ofthe residual block transformed/quantized by the transformer/quantizer 130. Herein, the scanning method considers the characteristics of a transform technique, a quantization technique, and a block (macroblock or subblock), and the scanning sequence may be determined so that the scanned quantized transform coefficient string has the minimum strength. Although FIG. 1 illustrates that the scanner 140 is implemented separately from the encoder 150, the scanner 140 may be omitted and its function may be integrated intő the encoder 150.

[0054] The encoder 150 generates encoded data by encoding a transformed/quantized residual block. Specifically, the encoder 150 generates encoded data by encoding a quantized transform coefficient string generated by scanning the quantized transform coefficients of a residual block transformed/quantized by the transformer/quantizer 130, or by encoding a quantized transform coefficient string generated by being scanned by the scanner 140.

[0055] An entropy encoding technology may be used as the encoding technology, although other unlimited encoding technologies may be used as the encoding technology. In addition, the encoder 150 may include nőt only a bitstream obtained by encoding a quantized transform coefficient string, bút alsó various information necessary to decode an encoded bitstream, in the encoded data. Herein, various information necessary to decode an encoded bitstream may include information on a block type, information on an intra prediction mode (ifthe prediction mode is an intra prediction mode), information on a motion vector (if the prediction mode is an inter prediction mode), information on a transform/quantization type, and various other information.

[0056] The inverse quantizer/transformer 160 reconstructs a residual block by inverse-quantizing/transforming a residual block transformed/quantized by the transformer/quantizer 130. The inverse quantization/transform may be earried out by inversely performing the transform/quantization process ofthe transformer/quantizer 130. That is, the inverse quantizer/transformer 150 may perform inverse quantization/transform by inversely performing the transform/quantization process ofthe transformer/quantizer 130 by using transform/quantization related information (e.g., information on a transform/quantization type) generated and transmitted by the transformer/quantizer 130.

[0057] The adder 170 reconstructs a current block by adding the predicted block predicted by the predictor 110 and the residual block inverse-quantized/transformed by the inverse quantizer/transformer 160.

[0058] The filter 180 filters the current block reconstructed by the adder 170. The filter 180 reduces a blocking effect that is generated at a block boundary or a transform boundary by transform/quantization of a videó in units of a block. The filter 180 may perform filtering by using transform/quantization type information transmitted togetherwith a reconstructed current block. The transform/quantization type information may be transmitted by the inverse quantizer/transformer 160 to the adder 170 and then transmitted to the filter 180.

[0059] Adeblocking filter is used to remove the blocking effect, which may be equivalent to a loop filter. The deblocking filter may alsó perform filtering on a boundary between blocks (which may be Μ x N macroblocks), a boundary between transforms according to a P x Q transform size in a macroblock determined by the transformer/quantizer 130, and a boundary between a block and a transform. A P x Q transform type refers to both a square transform type and a rectangular transform type, and a blocking effect is generated by performing transform/quantization according to a transform unit. In orderto remove the blocking effect, a deblocking filter may be applied to both a macroblock boundary and a transform boundary. Consequently, filtering may be applied to all boundaries according to a macroblock type and a transform type, so that a filtering process for removing a blocking effect can be applied. As for a filtering method for removing the blocking effect, an embodiment of the present disclosure is different from the conventional H.264/AVC in that it applies a filtering method according to a transform block size in consideration of all the PxQ transform boundaries.

EP 3 119 091 Β1 [0060] In the conventional H.264, a macroblock type used for videó encoding is in 16x16 pixels, and a predicted block is generated by performing intra/inter prediction on each macroblock. An encoding method for encoding a videó in units of a macroblock is widely used because it can encode a videó in consideration of the régiónál characteristics of the videó. In addition, sinee various intra/inter predictions are used for génération of a predicted block, a high encoding efficiency can be provided.

[0061] FIGs. 2 to 4 are diagrams illustrating intra prediction modes according to macroblock types used in typical videó encoding.

[0062] FIG. 2 is a diagram illustrating nine intra prediction modes in the case where a macroblock type is an intra 4x4 macroblock. FIG. 3 is a diagram illustrating nine intra prediction modes in the case where a macroblock type is an intra 8x8 macroblock. FIG. 4 is a diagram illustrating four intra prediction modes in the case where a macroblock type is an intra 16x16 macroblock.

[0063] In the case of intra prediction, an encoded adjacent block is used to generate a predicted block in units of a 4x4, 8x8 or 16x16 block as illustrated in FIG. 2. In the case of inter prediction, a previous-encoded frame is used to generate a predicted block in units of a16x16, 16x8, 8x16 or 8x8 block as illustrated in FIG. 3. If a predicted block is generated in units of an 8x8 block, each 8x8 block is used to generate a predicted block in units of an 8x8, 8x4, 4x8 or 4x4 block.

[0064] If a macroblock type is an intra block type, a macroblock to be encoded is predicted by intra prediction. The intra block type is subdivided intő an intra 4x4 macroblock, an intra 8x8 macroblock, and an intra 16x16 macroblock. In each case, a macroblock is predicted by using adjacent pixels of an adjacent block that is aíready encoded, decoded and reconstructed, according to a prediction mode illustrated in FIGs. 2 to 4.

[0065] FIG. 5 is a diagram illustrating an inter prediction mode according to a macroblock type used in typical videó encoding.

[0066] If a macroblock type is an inter block type, a macroblock to be encoded is predicted by inter prediction. In this case, as illustrated in FIG. 3, a macroblock is predicted in units of a 16x16, 16x8, 8x16 or 8x8 block by using a frame that is aíready encoded, decoded and reconstructed, to generate a predicted block. If a macroblock is predicted in units of an 8x8 block, each 8x8 block is predicted in units of an 8x8, 8x4, 4x8 or 4x4 block to generate a predicted block. [0067] In addition, H.264 uses a 4x4 or8x8 integertransform based on diserete cosinetransform. An integertransform does nőt perform a reál number-based operation that is a drawback of diserete cosine transform, and performs only an integer-based operation while maintaining, to the utmost, the characteristics of diserete cosine transform. Therefore, the integer transform is advantageous in terms of encoding efficiency and complexity. A deblocking filter is used to remove a blocking effect caused by a block-based transform.

[0068] However, in the case of encoding a high-resolution videó, using various PxQ transforms may be more efficient than using only a 4x4 or 8x8 transform used in H.264. Herein, if a P x Q transform is used, it is necessary to apply a deblocking filter in units ofa block as in the conventional H.264/AVC and alsó to apply a deblocking filter to a portion to be filtered in accordanee with various types of transform boundaries. In addition, in the case of filtering, as for a filtering method applied according to various macroblock or transform types, a more improved performance and videó quality can be achieved only when a deeper and stronger filtering is performed on a macroblock or transform boundary larger than that of the conventional method.

[0069] In an exemplary embodiment ofthe present disclosure, the filter 180 performs filtering on a transform boundary sízed nőt to be performed in H.264/AVC with respect to a P x Q transform applied to all or somé sizes of an Μ χ N macroblock. First, a transform type is identified through various information such as a bitstream or a block type with respect to a transform type determined by the transformer/quantizer 130, and a blocking-effect portion and a nonblocking-effect portion at a boundary between transforms are diseriminated from each other. This is to prevent an unnecessary filtering from being applied to the non-blocking-effect portion and to accurately detect information about the non-blocking-effect portion and the blocking-effect portion so that a filtering is applied to a transform boundary position at which a blocking effect is generated.

[0070] In an embodiment ofthe present disclosure, a boundary between transforms, a boundary between a block and a transform, and a boundary between blocks may be divided prior to filtering. Although only the removal of a blocking effect generated at a boundary between transforms is deseribed in the following deseription, this is merely for the convenience of deseription, and a filtering method for a boundary between blocks and a filtering method for a boundary between a transform and a block may be fundamentally similar to a filtering method for a boundary between transforms. [0071] In a filtering proeess, the filter 180 determines a transform boundary, pixels adjacent to the boundary to be filtered, and a filtering strength. Herein, the filtering is performed áfter determining whether the pixels adjacent to the boundary correspond to an actual edge of a videó or a blocking region generated by block transform/quantization. [0072] In embodiments ofthe present disclosure, a filtering strength has the same meaning as a boundary strength (BS). [0073] FIG. 6 is a diagram illustrating a method of determining an Μ χ N macroblock type and transform type according to an embodiment ofthe present disclosure, and FIG. 7 is a diagram illustrating a proeess ofidentifying a filtering boundary and strength according to an embodiment ofthe present disclosure.

ΕΡ 3 119 091 Β1 [0074] As illustrated in FIG. 6, a 16x16 macroblock may be divided intő two 16x8 subblocks. Herein, a size of the macroblock is nőt limited to 16x16, bút may be extended to Μ x N. In addition, a P x Q transform may be performed according to a block type. FIG. 6 illustrates two examples ofa transform/quantization type (which may be a transform size) for a 16x8 block. Specifically, a transform for a 16x8 block may include one 16x8 transform or two 8x8 transforms. Herein, when the transform having the least information amount among the transforms of two sizes is used, the highest encoding efficiency can be achieved.

[0075] Selection ofa transform type is described in detail. For example, if a block has a size of 16x8, a 16x8 transform may be used. Ifthe 16x8 transform is divided intő two 8x8 transforms, the transform with the best performance may be used. When a transform size is divided, a method of using a coded block pattern (CBP) value or inserting other information intő a bitstream may be used and a method of determining it by a block type may alsó be used. In order to select an optimál transform for a block to be encoded, rate-distortion (RD) costs for a plurality of transform types ofthe block may be sequentially calculated, and a transform size with the least RD cost may be selected as a final transform size. For example, RD costs for a 16x8 transform, an 8x8 transform, an 8x4 transform, a 4x8 transform and a 4x4 transform ofa 16x8 block may be sequentially calculated, and a transform size with the least RD cost may be selected as a final transform size.

[0076] Information on the transform size determined as above is used to perform deblocking filtering in accordance with a transform boundary. Since a blocking effect is generated differently according to transform sizes, a blocking boundary is identified prior to filtering. Whether the blocking boundary is an actual transform boundary may be identified by using information on a bitstream of a reconstructed block or other types of information on a transform included in a reconstructed block, prior to deblocking filtering. Herein, the identification is to set different boundary strengths.

[0077] According to an embodiment ofthe present disclosure, boundaries between transform blocks in a picture to be filtered (that is, transform boundaries) are identified first. When the transform boundaries are identified, a filtering strength and pixels of a filtering boundary region are determined in consideration of at least a size of two adjacent transform blocks forming each boundary.

[0078] Referring to FIG. 7, it is determined whether a boundary between a region A and a region B is a transform boundary or an edge ofan actual image (S710). Ifthe boundary corresponds to an edge ofan actual image, a BS value is set to 0 and the boundary is nőt filtered. If nőt, the boundary is identified as a portion from which a blocking effect is removed. Thereafter, if filtering is necessary, it is determined whetherthe boundary is a large transform boundary (S720). In a region where filtering is performed on all boundaries according to a block size and transform, it is difficult to remove a blocking effect on a large transform or a block boundary by a conventional filtering strength (or boundary strength). This is because H.264/AVC performs encoding in units of a 16x16 macroblock to the maximum, and a transform size is unsuitable torán MxN block because it has a square transform. Herein, whether it is a transform boundary is identified in determining a filtering strength (S710). It is determined whether a large-block square or rectangular transform, which was nőt used in the conventional H.264/AVC, is applied (S720). If a large-block square or rectangular transform is applied, a filtering stronger than a conventional filtering (BS = 5) is performed.

[0079] Herein, the large-block square or rectangular transform is more than 16x16, and in the case of a square or a rectangle, it can be applied to have the highest boundary strength. In somé cases, a 16x8 or larger size or an 8x16 or larger size may be a reference size, and whether it is a large-size transform may be determined by various transform sizes; however, the embodiment of the present disclosure is nőt limited thereto. If it is nőt a large transform, a filtering strength determining process (S730 to S780) is similar to that of the conventional H.264/AVC, and thus a detailed description thereof will be omitted in order nőt to obscure the subject matters of the present disclosure. In addition, the embodiment ofthe present disclosure is nőt limited to the method illustrated in FIG. 7, and the transform boundary may be identified through other methods. Even when other methods are used, the transform boundary can be identified to have the same effect.

[0080] FIG. 8 is a diagram illustrating an example of a filtering method for large-block transform according to an embodiment of the present disclosure.

[0081] As illustrated in FIG. 8, if a rectangular transform is performed, up/down/left/right boundaries may have different types. For example, transform boundaries such as 16x16, 16x8 and 8x4 are illustrated in FIG. 8. In addition, a method of performing filtering by the filter 180 according to an embodiment ofthe present disclosure can alsó besimilarly applied to 16x16, 16x8 and 8x4 transforms or extended P x Q transforms as illustrated in FIG. 8, as well as to 8x8 and 4x4 transforms. As for a boundary of a large block, if it is determined that the boundary is nőt an edge of an actual image, a blocking effect can be effectively removed by applying a deblocking filter to more pixels. A technology of determining whether a transform boundary is an edge of an actual image from a transform (block) boundary pixel is well known in the art, and thus a detailed description thereof will be omitted.

[0082] Hereinafter, an embodiment of the present disclosure in which a filtering strength or pixels to be filtered are selected according to a type or size of a transform used will be described in detail with reference to FIG. 8.

[0083] As described above, in the conventional H.264/AVC standard, only a 4x4 transform and an 8x8 transform are used, and a rectangular transform or a larger transform is nőt used. An embodiment ofthe present disclosure provídes

ΕΡ 3 119 091 Β1 block boundary filtering or transform boundary filtering that is suitable for the case of using a rectangular transform or a transform larger than a conventional transform.

[0084] FIG. 8 illustrates two reproduced 16x16 macroblocks MB1 and MB2 prior to filtering. The left macroblock MB1 includes a 16x8 block SMBO encoded and/or decided by using a 16x8 transform, and four 8x4 blocks SMB1, SMB2, SMB3 and SMB4 encoded and/ordecoded by using a 8x4 transform. The right macroblock MB2 uses a 16x16 transform. [0085] Since SMBO is adjacent to MB2 in the horizontal direction, a boundary thereof is forrned in the verticaí direction. In addition, since SMBO is adjacent to SMB1 and SMB2 in the verticaí direction, a boundary thereof is forrned in the horizontal direction.

[0086] A filter 960 of a videó decoding apparátus 900 or the filter 180 of the videó encoding apparátus 100 according to an embodiment of the present disclosure determines the positíon and/or the number of pixels in a transform block to be filtered, according to the size of at least one transform block or the sizes of two transform blocks forming a transform boundary or a boundary between transform blocks.

[0087] The filter 180/960 identifies the sizes of SMBO, SMB1, SMB2 and MB2 in order to filter a verticaí boundary between SMBO and MB2 and a horizontal boundary between SMBO and SMb1 or SMB2. In particular, according to an embodiment of the present disclosure, the horizontal size (length) of SMBO and MB2 is considered in order to filter the verticaí boundary, and the verticaí size (length) of SMBO, SMB1 and SMB2 is considered in order to filter the horizontal boundary.

[0088] Referring to FIG. 8, the horizontal length of SMBO and MB2, that is, the number of pixels of the two blocks in the horizontal direction is 16. The verticaí length of SMBO, that is, the number of pixels of SMBO in the verticaí direction is 8, and the number of pixels of SMB1 and SMB2 in the verticaí direction is 4.

[0089] If a verticaí boundary region between SMBO and MB2 is filtered, the filter 180/960 filters six pixels qO, q1, q2, q3, q4 and q5 in SMBO that are continuous in the horizontal direction from the verticaí boundary, and filters six pixels pO, p1, p2, p3, p4 and p5 in MB2. On the other hand, in order to remove a block distortion present at the horizontal boundary between SMBO and SMB1, the filter 180/960 filters three pixels pO, p1 and p3 in SMB1 that are continuous in the verticaí direction from the horizontal boundary, and filters three pixels qO, q1 and q2 in SMBO. That is, as the size of a transform block increases, more pixels are filtered.

[0090] The filter 180/960 determines the positíon and/or the number of pixels in a transform block to be filtered, according to the size of at least one transform block or the sizes of two transform blocks forming a boundary between transform blocks. Filtering up to which pixel away from the boundary may be determined by the smaller block among the two transform blocks.

[0091] The videó encoding apparátus 100 or the videó decoding apparátus 900 may know a transform block size in various ways. For example, a transform block size may be indicated by a syntax element of a bitstream representing a block type íncluding a transform block size óra size ofa transform used in encoding or decoding a relevant transform block. [0092] According to another embodiment ofthe present disclosure, the filter 180 ofthe videó encoding apparátus 100 or the filter 960 of the videó decoding apparátus 900 determines a filtering strength according to the size of at least one transform block or the sizes of two transform blocks forming a boundary between transform blocks.

[0093] As deseribed above, the filter 180/960 identifies the sizes of SMBO, SMB1, SMB2 and MB2 in order to filter a verticaí boundary between SMBO and MB2 and a horizontal boundary between SMBO and SMb1 or SMB2. In particular, according to an embodiment of the present disclosure, the horizontal size (length) of SMBO and MB2 is considered for a filtering strength of the verticaí boundary, and the verticaí size (length) of SMBO, SMB1 and SMB2 is considered for a filtering strength ofthe horizontal boundary.

[0094] In the present embodiment, a filtering strength for a pixel in a block with a larger transform block size is higher than a filtering strength for a pixel in a block with a smaller transform block size. Thus, a filtering strength of the verticaí boundary between SMBO and Mb2 is greater than a filtering strength ofthe horizontal boundary between SMBO and SMB1 and between SMBO and SMB2.

[0095] When the filtering strength and the pixels to be filtered are determined as deseribed above, the pixels of a block boundary region are filtered according to the filtering strength determined. A filtering method according to a filtering strength (BS) according to an embodiment of the present disclosure will be deseribed below in detail. If the size of BS is smaller than 4, a filtering process is performed as Equation 1; and if the size of BS is 4, a filtering process is performed as Equation 2.

ΕΡ 3 119 091 Β1 {(q0 — ρθ) « 2 + (pl — ql) + 4}

Δ = Clip —te, te, p'O = ρθ + Δ q'O = q'0 + Δ

Equation 1 [0096] Equation 1 is an exampie of when a BS is smaller than 4, and te is determined by |p2-p0|, |q2-q0| and β determined by a quantization index. In the case of Clip[a, b, c}, a Clip function is processed such that c is between a and b. That is, Clip of Equation 1 is processed such that ((q0 - p0)«2 + (p1 - q1) +4)/8 is between -te and te. As expressed in Equation 1, p’O and q’O may be obtained through 4-tap filtering using q1, qO, pO and p1. Filtering of pixel values p’1 and q’1 may be performed through a method similar to a method of obtaining p’O and q’O.

<img file="HUE035559T2_D0001.tif" />

lxq2 + 2xql + 2xq0 + 2xp0 + lxpl + 4

Equation 2 [0097] Equation 2 is an exampie of an equation for obtaining q’O in the case of BS=4. Herein, 5-tap filtering is applied for a filtering coefficient value in the order of 1, 2, 2, 2, 1, and whether it is an actual edge is identified by a and β determined by a quantization index. Filtering applied to other pixels generates p’2~q’2 by using a method similar to the conventional H.264/AVC. Since the maximum number of pixels filtered is limited to 6 (6 for a luminance signal, and 4 for a chrominance signal), up to 3 pixels can be filtered accordíng to the conventional method.

[0098] An embodiment of the present disclosure provides a filtering mode further including a boundary strength (for exampie, BS is called 5) in the case where a boundary strength varies accordíng as a block size increases or a transform size varies. That is, as in the case of BS being 5, in the case of a large block transform, filtering illustrated in FIG. 8 is performed such that a deeper pixel is influenced. Since the number of conventional filtering pixels is nőt suitable for a block transform larger than a transform applied in the conventional filtering method, a blocking effect cannot be effectively redueed. Therefore, in an embodiment of the present disclosure, the number of conventional filtering pixels and the conventional filtering method are improved in orderto solve the above problem.

[0099] Equation 3 is an exampie ofa filtering method having more pixels.

q'O = lxq3 + 2xq2 + 3xql+4xq0 + 3xp0 + 2xpl + lxp2 + 8

Equation 3 [0100] Equation 3 is an equation for obtaining q’O in a large block transform. Since the number of pixels involved in filtering is increased as compared to the conventional method and 7-tap filtering of 1,2, 3, 4, 3, 2, 1 is performed, it is more influenced by adjacent pixels. Since a large block is filtered such that it is influenced by more adjacent pixels, a blocking effect can be redueed more effectively.

q'4 =

4xq5 + 3xq4 + 3xq3+2xq2 + 2xql + lxqO + lxpO + 8

Equation 4 [0101] Equation 4 is an exampie of filtering a pixel q’4. In the case ofa large transform size, since the pixel q’4 that is nőt filtered in the conventional technology is filtered, a videó quality can be improved. This is to vary the number of pixels involved in filtering as a block size varies. In the embodiment ofthe present disclosure, a filtering method for the number of pixels to be filtered is nőt limited to the methods illustrated in the above equations, and various other filtering methods may be applied differently accordíng to block types and transform types.

[0102] The embodiment of the present disclosure improves a filtering method for a transform boundary or a block boundary that can alsó be effectively applied to high-quality videó coding to be developed in the future, by solving the problem of a conventional filtering method for a P x Q transform of an Μ x N block with respect to a deblocking filter that is a method for removing a conventional blocking effect. Accordíng to the embodiment ofthe present disclosure, filtering can be performed very effectively in removing a blocking effect at a transform boundary and a macroblock of every size. [0103] As deseribed above, when the videó encoding apparátus 100 accordíng to an embodiment of the present disclosure and the videó encoding method using the same are used, a blocking effect can be removed more effectively

ΕΡ 3 119 091 Β1 by using information on an Μ χ N transform, a P x Q transform, and a transform type. In this manner, the videó encoded with encoded data by the videó encoding apparátus 100 may be transmitted in reál time or non-real-time to the videó decoding apparátus to be described later where it is reconstructed and reproduced intő the videó after being transmitted via a wired/wireless communication network including the Internet, a short rangé wireless communication network, a wireless LAN network, a WiBro (Wireless Broadband, alsó known asWiMax) network, and mobile communication network or a communication interface such as cable or USB (universal serial bús).

[0104] FIG. 9 is a block diagram illustrating a schematic configuration of a videó decoding apparátus according to an embodiment of the present disclosure.

[0105] Like the videó encoding apparátus 100 described with reference to FIG. 1, a videó decoding apparátus 900 according to an embodiment of the present disclosure may be a personal computer (PC), a notebook computer, a television (TV), a personal digital assistant (PDA), a portable multimédia player (PMP), a Playstation Portable (PSP), a mobile communication terminál, óra digital TV, and may represent a variety of apparatuses equipped with, for example, a communication device such as a modem for carrying out Communications between various devices or wired/wireless communication networks, a memory for storing data and various programs for encoding images or videós, and a microprocessor for executing the programs to effect operations and Controls.

[0106] Avideo decoding apparátus 900 according to an embodiment ofthe present disclosure may inelude a decoder 910, an inverse scanner 920, an inverse quantizer/transformer 930, a predictor 940, an adder 950, and a filter 960. Herein, the inverse scanner 920 and the filter 960 are nőt necessarily included bút may be omitted selectively according to implementation modes. If the inverse scanner 920 is omitted, a function ofthe inverse scanner 920 may be integrated intő the decoder 910.

[0107] The decoder 910 reconstructs a transformed/quantized residual block by decoding encoded data. Specifically, the decoder 910 reconstructs a quantized transform coefficient string by decoding encoded data. If a function of the scanner 140 is integrated intő the encoder 150 in the videó encoding apparátus 100, the inverse scanner 920 is omitted in the videó decoding apparátus 900 and a function of the inverse scanner 920 is integrated intő the decoder 910. Therefore, the decoder 910 may reconstruct a transformed/quantized residual block by inverse-scanning a reconstructed quantization coefficient string.

[0108] In addition, the decoder 910 may decode nőt only a transformed/quantized residual block bút alsó information necessary for decoding by decoding encoded data. The information necessary for decoding refers to information necessary to decode an encoded bitstream in encoded data, and may inelude information on a block type, information on an intra prediction mode (if the prediction mode is an intra prediction mode), information on a motion vector (if the prediction mode is an inter prediction mode), information on a transform/quantization type, and various other information. [0109] The information on a block type may be transmitted to the inverse quantizer/transformer 930 and the predictor 940. The information on a transform/quantization type may be transmitted to the inverse quantizer/transformer 930. Information necessary for prediction such as the information on an intra prediction mode and the information on a motion vector may be transmitted to the predictor 940.

[0110] When the decoder 910 reconstructs and transmits a quantized transform coefficient string, the inverse scanner 920 reconstructs a transformed/quantized residual block by inverse-scanning the quantized transform coefficient string. [0111] The inverse scanner 920 generates a residual block with a quantization coefficient by inverse-scanning an extracted quantization coefficient string by various inverse scanning methods such as inverse zigzag scanning. Herein, information about a transform size is obtained from the decoder 910, and an inverse scanning method corresponding to the information is used to generate a residual block.

[0112] In addition, as described above, if a function of the scanner 140 is integrated in the encoder 150 in the videó encoding apparátus 100, the inverse scanner 920 may alsó be omitted in the videó decoding apparátus 900 and a function ofthe inverse scanner 140 may be integrated intő the decoder 910. In addition, the decoder 910 or the inverse scanner 920 inverse-scans a transformed/quantized residual block according to a transform/quantization type identified by information on a transform/quantization type reconstructed by decoding encoded data by the decoder 910. Herein, since an inverse scanning method performed by the inverse scanner 920 according to the transform/quantization type is identical toorsimilartoan inversion ofthe method ofscanning quantization transform coefficients of a transformed/quantized residual block by the scanner 140, a detailed description ofthe inverse scanning method will be omitted.

[0113] The inverse quantizer/transformer 930 reconstructs a residual block by inverse-quantizing/transforming a reconstructed transformed/quantized residual block. Herein, the inverse quantizer/transformer 930 inverse-quantizes/transforms a transformed/quantized residual block according to a transform/quantization type identified by information on a transform/quantization type received from the decoder 910. Herein, since a method of inverse-quantizing/transforming a transformed/quantized residual block by the inverse quantizer/transformer 930 according to a transform/quantization type is identical to or similar to an inversion of the transform/quantization process performed by the transformer/quantizer 130 of the videó encoding apparátus 100 according to a transform/quantization type, a detailed description ofthe inverse quantization/transform method will be omitted.

[0114] The predictor 940 generates a predicted block by predicting a current block. Herein, the predictor 940 predicts

EP 3 119 091 Β1 the current block by using information necessary for prediction and information on a block type received from the decoder 910. That is, the predictor 940 generates a predicted block by determining a size and type ofthe current block according to a block type identified by information on a block type and predicting a current block by using a motion vector or an intra prediction mode identified by information necessary for prediction. Herein, the predictor 940 may generate the predicted block by eombining predicted subblocks generated by dividing the current block intő subblocks and predicting the respective subblocks.

[0115] The adder 950 reconstruct the current block by adding the residual block reconstructed by the inverse quantizer/transformer 930 and the predicted block generated by the predictor 940.

[0116] The filter 960 filters the current block reconstructed by the adder 950, and the current block reconstructed and filtered is accumulated in units ofa picture and stored as a reference picture in a memory (nőt illustrated) and used by the predictor 940 to predict a next block or a next picture.

[0117] Here, in the filtering, different filtering boundary strengths may be determined with respect to a boundary between transforms, a boundary between a block and a transform, and a boundary between blocks.

[0118] In addition, the filtering may be performed in the case of no edge in the boundary and the boundary strength may be determined according to a transform and quantization type. Herein, if the transform/quantization type is larger than 16x8 or 8x16, the greatest boundary strength may be provided.

[0119] As a transform/quantization size increases, the number of pixels involved in filtering may increase. The filter 960 may perform filtering by using information about a transform/quantization type transmitted together with a reconstructed current block.

[0120] When filtering a reconstructed current block, the filter 960 may perform the filtering according to a transform/quantization type identified by information on a transform/quantization type received from the decoder 910. Herein, the filter 960 may perform a deblocking filtering on a transform boundary or a block boundary differently according to transform/quantization types, to reduce a blocking effect generated at a block boundary of an image. Since a filtering method ofthe filter 960 is identical to or similar to the deblocking filtering process performed by the filter 180 ofthe videó encoding apparátus 100, a detailed description ofthe filtering method will be omitted.

[0121] A videó encoding/decoding apparátus according to an embodiment of the present disclosure may be implemented by eombining the videó encoding apparátus 100 of FIG. 1 and the videó decoding apparátus 900 of FIG. 9. [0122] A videó encoding/decoding apparátus according to an embodiment of the present disclosure may include a videó encoder 100 (that is, a videó encoding unit in the videó encoding/decoding apparátus according to an embodiment of the present disclosure) and a videó decoder 900 (that is, a videó decoding unit in the videó encoding/decoding apparátus according to an embodiment of the present disclosure). The videó encoder 100 may include a predictor 110 for generating a predicted block by predicting a current block, a subtractor 120 for generating a residual block by subtracting the predicted block from the current block, a transformer/quantizer 130 for determining a transform/quantization type seleeted according to a block type ofthe current block and transforming/quantizing the residual block according to the transform/quantization type determined; an encoder 150 for generating encoded videó data by encoding a transformed/quantized residual block; an inverse quantizer/transformer 160 for reconstructing a residual block by inversequantizing/transforming the transformed/quantized residual block; an adder 170 for generating a reconstructed block by adding the predicted block to the residual block reconstructed; and a filter 180 for filtering the reconstructed block according to the transform/quantization type. The videó decoder 900 may include a decoder 910 for reconstructing a transformed/quantized residual block by decoding encoded data; an inverse quantizer/transformer 930 for reconstructing the residual block by inverse-quantizing/transforming the transformed/quantized residual block according to a transform/quantization type; a predictor 940 for generating a predicted block by predicting a current block; an adder 950 for reconstructing the current block by adding a reconstructed residual block and the predicted block; and a filter 960 for filtering a boundary region between a reconstructed current block and an adjacent block according to the transform/quantization type.

[0123] Here, in the filtering method for reducing a videó data quantization effect ofthe videó encoding apparátus 100 and the videó decoding apparátus 900, the filtering method of the filter 180 and the filter 960 in the videó encoding apparátus 100 and the videó decoding apparátus 900 performs filtering in the horízontal direction and then in the vertical direction or performs filtering in the vertical direction and then in the horízontal direction, in order to prevent a value mismatch after the filtering operation ofthe videó encoding apparátus 100 and the videó decoding apparátus 900. This deblocking filtering sequence may be determined as the same sequence in the videó encoding apparátus 100 and the videó decoding apparátus 900.

[0124] FIG. 10 is a flow diagram illustrating a videó encoding method according to an embodiment ofthe present disclosure.

[0125] A videó encoding method according to an embodiment of the present disclosure may include: generating a predicted block by predicting a current block (1002); generating a residual block by subtracting the predicted block from the current block (1004); determining a transform/quantization type seleeted according to a block type of the current block (1006); transforming/quantizing the residual block according to the transform/quantization type determined (1008);

EP 3 119 091 Β1 generating encoded videó data by encoding a transformed/quantized residual block (1010); reconstructing a residual block by inverse-quantizing/transforming the transformed/quantized residual block (1012); generating a reconstructed block by adding the predicted block to a reconstructed residual block (1014); and filtering a boundary region between the reconstructed block and an adjacent block aecording to the transform/quantization type (1016).

[0126] Herein, the filtering may determine different filtering boundary strengths with respect to a boundary between transforms, a boundary between a block and a transform, and a boundary between blocks.

[0127] Herein, the filtering may be performed in the case of no edge in the boundary, and the boundary strength may be determined aecording to a transform and quantization type.

[0128] In addition, if the transform/quantization type is larger than 16x8 or 8x16, the greatest boundary strength may be provided.

[0129] In the filtering, boundary pixels involved in the filtering may be determined aecording to a transform size, and the number of pixels involved in the filtering may increase with an increase in the transform/quantization size.

[0130] In the transform/quantization, information on a transform/quantization type may be generated.

[0131] Rate-distortion (RD) costs for a plurality of transform types of the residual block may be calculated, and a transform type with the Ieast RD cost calculated may be selected as a transform/quantization type.

[0132] The transform type may be P x Q (P and Q may be different from each other), and P and Q may be equal to or larger than 16. That is, a size of one side of a block may be equal to or larger than 16.

[0133] The filtering may be performed by using information on a transform/quantization type transmitted together with a reconstructed block.

[0134] The filtering may be performed after it is determined whether a boundary between the reconstructed block and the adjacent block is a transform boundary. If the boundary is a transform boundary, the boundary region may be filtered; and if the boundary does nőt correspond to an edge of an image, the filtering may be performed.

[0135] In the transform/quantization, information on a transform/quantization type may be generated.

[0136] Herein, the filtering may be performed in the same order as in a videó decoding method.

[0137] FIG. 11 is a flow diagram illustrating a videó decoding method aecording to an embodiment ofthe present disclosure.

[0138] A videó decoding method aecording to an embodiment of the present disclosure may include: outputting a transformed/quantized residual block by decoding encoded data (1102); decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual block aecording to a transform/quantization type (1104); generating a predicted block by predicting a current block (1106); reconstructing the current block by adding a decoded residual block and the predicted block (1108); and filtering a reconstructed current block aecording to the transform/quantization type (1110).

[0139] The filtering may determine different filtering boundary strengths with respect to a boundary between transforms, a boundary between a block and a transform, and a boundary between blocks.

[0140] The filtering may be performed after it is determined whether a boundary between the reconstructed current block and the adjacent block is a transform boundary. If the boundary is a transform boundary, the boundary region may be filtered; and if the boundary does nőt correspond to an edge of an image, the filtering may be performed.

[0141] Herein, the filtering may be performed in the case of no edge in the boundary, and the boundary strength may be determined aecording to a transform and quantization type.

[0142] In addition, if the transform/quantization type is larger than 16x8 or 8x16, the greatest boundary strength may be provided.

[0143] In the filtering, boundary pixels involved in the filtering may be determined aecording to a transform size, and the number of pixels involved in the filtering may increase with an increase in the transform/quantization size.

[0144] The filtering may be selected aecording information of a transform type included in the encoded data among a plurality of transform types. That is, information on a transform/quantization type may be generated in the inverse quantization/transform after the decoding ofthe encoded data, and the filtering may be performed by using information on a transform/quantization type transmitted together with the reconstructed current block.

[0145] The filtering may be performed in the same order as in the videó encoding method.

[0146] A videó encoding/decoding method aecording to an embodiment ofthe present disclosure may be implemented by combining a videó encoding method aecording to an embodiment of the present disclosure and a videó decoding method aecording to an embodiment ofthe present disclosure.

[0147] Avideo encoding/decoding method aecording to an embodiment ofthe present disclosure may include: a videó encoding method aecording to an embodiment of the present disclosure for generating a predicted block by predicting a current block, generating a residual block by subtracting the predicted block from the current block, determining a transform/quantization type selected aecording to a block type ofthe current block, transforming/quantizing the residual block aecording to the transform/quantization type determined, generating encoded videó data by encoding a transformed/quantized residual block, reconstructing a residual block by inverse-quantizing/transforming the transformed/quantized residual block, generating a reconstructed block by adding the predicted block to a reconstructed

ΕΡ 3 119 091 Β1 residual block, and filtering the reconstructed block according to the transform/quantization type; and a videó decoding method according to an ernbodiment of the present disclosure for outputting a transformed/quantized residual block by decoding encoded data, decoding a residual block by inverse-quantizing/transforming the transformed/quantized residual blockaccordingto a transform/quantization type, generating a predicted block by predicting a current block, reconstructing the current block by adding a decoded residual block and the predicted block, and filtering a boundary ofa reconstructed current block according to the transform/quantization type.

[0148] FIG. 12 is a block diagram illustrating an image filtering apparátus according to an ernbodiment ofthe present disclosure.

[0149] An image filtering apparátus according to an ernbodiment ofthe present disclosure may include a boundary identifier 1210, a pixel/strength selecting unit 1220, and a filtering unit 1230.

[0150] The boundary identifier 1210 identifies a boundary between two transform blocks included in an image. [0151] The pixel/strength selecting unit 1220 selects one or more pixels to be filtered, according to a size of at Ieast one transform block among the two transform blocks.

[0152] The filtering unit 1230 filters at Ieast one pixel included in a region adjacent to the boundary.

[0153] In another ernbodiment, the pixel/strength selecting unit 1220 may select a filtering strength according to a size of at Ieast one transform block among the two transform blocks.

[0154] The image filtering apparátus according to an ernbodiment ofthe present disclosure may be used as the filter 180 in the videó encoding apparátus according to an ernbodiment of the present disclosure or the filter 960 in the videó decoding apparátus according to an ernbodiment ofthe present disclosure.

[0155] The boundary identifier 1210 may be used to identify a boundary between two transform blocks included in an image in the filter 180/960. That is, the boundary identifier 1210 may be used to identify a boundary region between a current block and an adjacent block. Herein, although identifying a boundary between a reconstructed block and an adjacent block is exemplified, the present disclosure is nőt limited thereto and it may be used forany purpose of identifying a boundary between two blocks. Since a method of identifying a boundary between two blocks has been described above with reference to the drawings including FIG. 8 and the above equations, a detaiied description thereof will be omitted herein.

[0156] The pixel/strength selecting unit 1220 may select a pixel to be filtered according to asizeof at Ieast one transform block among the two blocks, or may select a filtering strength according to a size of at Ieast one transform block among the two blocks. Since a method of selecting a pixel to be filtered according to a size of a transform block or selecting a filtering strength according to a size of a transform block has been described above with reference to the drawings including FIG. 8 and the above equations, a detaiied description thereof will be omitted herein.

[0157] The filtering unit 1230 filters at Ieast one pixel included in a region adjacent to the boundary. Since the filtering method has been described above with reference to the drawings including FIG. 8 and the above equations, a detaiied description thereof will be omitted herein.

[0158] The size of at Ieast one transform block may be a length of at Ieast one transform block in a perpendicular direction with respect to the boundary.

[0159] In addition, the size of at Ieast one transform block may be proportional to the number of pixels to be filtered. The size of at Ieast one transform block may be proportional to the filtering strength.

[0160] FIG. 13 is a flow diagram illustrating an image filtering method according to an ernbodiment of the present disclosure.

[0161] First, a boundary between two transform blocks included in an image is identified (S1302).

[0162] Thereafter, a pixel to be filtered (and/or a filtering strength) is selected according to a size of at Ieast one transform block among two transform blocks (S1304), and a pixel included in a region adjacent to the boundary is filtered (S1306).

[0163] The method of step S1302 may be used as a method for identifying a boundary between two transform blocks included in an image in the filter 180/960. That is, the boundary identifier 1210 may be used to identify a boundary region between a current block and an adjacent block. Herein, although identifying a boundary between a reconstructed block and an adjacent block is exemplified, the present disclosure is nőt limited thereto and it may be used for any purpose of identifying a boundary between two blocks. Since a method of identifying a boundary between two blocks has been described above with reference to the drawings including FIG. 8 and the above equations, a detaiied description thereof will be omitted herein.

[0164] In step S1304, a pixel to be filtered (and/or a filtering strength) may be selected according to a size of at Ieast one transform block among two transform blocks. Since a method of selecting a pixel to be filtered according to a size of a transform block or selecting a filtering strength according to a size of a transform block has been described above with reference to the drawings including FIG. 8 and the above equations, a detaiied description thereof will be omitted herein.

[0165] In step S1306, at Ieast one pixel included in a region adjacent to the boundary is filtered. Since the filtering method has been described above with reference to the drawings including FIG. 8 and the above equations, a detaiied

EP 3 119 091 Β1 description thereof wiil be omitted herein.

[0166] Herein, the size ofat least one transform block may be a length ofat least one transform block in a perpendicular direetion with respect to the boundary.

[0167] In addition, the size of at least one transform block may be proportional to the number of pixels to be filtered. The size of at least one transform block may be proportional to the filtering strength.

[0168] In the description above, although all ofthe components ofthe embodiments ofthe present disclosure may have been described as assembled or operatively connected as a unit, the present disclosure is nőt intended to limit itself to such embodiments. Rather, within the objective scope of the present disclosure, the respective components may be selectively and operatively combined in any numbers. Every one ofthe components may alsó be implemented by itself in hardware while the respective ones can be combined in part or as a whole selectively and implemented in a computer program having program modules for executing functions ofthe hardware equivalents. Codesor code segments to constitute such a program may be easily deduced by a person skilled in the art. The computer program may be stored in computer readable média, which in operation can realize the embodiments of the present disclosure. The computer readable média may include magnetic recording média, optical recording média, and carrier wave média.

[0169] In addition, terms like ’include’, 'comprise’, and ’have’ should be interpreted in default as inclusive or open rather than exclusive or closed unless expressíy defined to the contrary. All the terms that are technical, scientific or otherwise agree with the meanings as understood by a person skilled in the art unless defined to the contrary. Common terms as found in dictionaries should be interpreted in the context of the related technical writings nőt too ideally or impractically unless the present disclosure expressíy defines them so.

[0170] Although exempiary embodiments of the present disclosure have been described for illustrative purposes, those skilled in the art wiil appreciate that various modifications, additions and substitutions are possíbie, without departing from essential characteristics ofthe disclosure. Therefore, exempiary embodiments ofthe present disclosure have nőt been described for limiting purposes. Accordingly, the scope of the disclosure is nőt to be limited by the above embodiments bút by the claims and the equivalents thereof.

[Industrial Applicability] [0171] As described above, the present disclosure can be applied to the technology of encoding/decoding a videó in various block sizes and transform sizes, to reduce a blocking effect that is caused by lossy compression through transform/quantization, and is highly useful in a videó encoding/decoding apparátus requiring an improved videó quality.