Video Decoding Method Using Adaptive Quantization Matrices

Video decoding procedure that uses adaptive quantization matrices.

Technical field

The present invention relates to a method of encoding moving images to encode moving images and generating continuous streams, and a method of decoding moving images to decode said coded continuous streams, as well as continuous streams.

Prior art

In the era of multimedia that comprehensively treats audio, video and other pixel values, current news media, that is, newspapers, magazines, television, radio, telephone and other media through which information is transmitted to people have recently been included within the scope of multimedia. In general, the term multimedia refers to something that is represented by associating not only characters, but also graphics, audio and, in particular, images and the like. However, a prerequisite for encompassing the available information media mentioned above in the field of multimedia is the representation of such information in digital format.

However, when the amount of information contained in each of the information media mentioned above is calculated as the amount of digital information, while the amount of information per character is 1 to 2 bytes in the case of the characters, the amount of information that is going to be needed in the case of audio (telephone quality) is 64 Kbits per second and, in the case of moving images (current television reception quality), 100 Mbits per second. Therefore, it is not reasonable for the aforementioned media to handle such an enormous amount of information as is available in digital format. For example, although there are currently video telephones that use the integrated services digital network (ISDN) that offers a transmission rate between 64 Kbits / s and 1.5 Mbits, it is not practical to transmit video from televisions and cameras directly through of ISDN.

From this point of view, the need for information compression techniques has been raised, and motion image compression techniques that comply with H.261 and H.263 standards recommended by ITU-T (Union) have been used international telecommunications, telecommunication standardization sector) for video telephones, for example. In addition, according to the information compression techniques that comply with the MPEG-1 standard, it is possible to store image information on a CD (compact disc) of current music along with sound information.

The term MPEG (Group of experts in motion images) refers to an international standard for compression of motion picture signals standardized by ISO / IEC (International Standardization Body, International Electrotechnical Commission), and the term MPEG-1 refers to to a standard for compressing television signal information to approximately one hundredth, so that motion picture signals can be transmitted at a speed of 1.5 Mbit / s. In addition, since the transmission speed achieved by the MPEG-1 standard is an average quality speed of about 1.5 Mbit / s, the MPEG-2 standard created with a view to meeting better image quality requirements allows transmission of data equivalent in quality to television broadcasting, whereby the motion picture signals are transmitted at a speed of 2 to 15 Mbit / s. On the other hand, the working group (ISO / IEC JTC1 / SC29 / WG11), promoter of the MPEG-1 and MPEG-2 standards, created the MPEG-4 standard. The MPEG-4 standard, which offers a higher compression rate than the MPEG-1 and MPEG-2 standards and offers object-based coding, decoding and operation, is capable of providing the new functions required in the Multimedia era. In the initial phase of normalization, the MPEG-4 standard was aimed at offering a low bit rate coding procedure, but later it has been extended to be a standard that supports more general coding that processes interlaced images, as well as coding. High bit rate. Currently, ISO / IEC and ITU-T have joined forces to standardize MPEG-4 AVC and ITU-T H.264 as next-generation image coding procedures that offer a higher compression rate. As of August 2002, a draft committee (CD) is issued for a next-generation image coding procedure.

In general, in the encoding of a moving image, the amount of information is compressed by reducing redundancies in the temporal and spatial directions. Therefore, in the interimage prediction coding that aims to reduce temporal redundancies, the estimation of the movement and the generation of a block-block predictive image with reference to the progressive or regressive images are carried out, and then the coding at the difference value between the predictive image obtained and an image of the current image to be encoded. Although the latter refers to an image, in the case of a progressive image "image" means "frame", while in the case of an interlaced image it means "frame" or "fields". Here, an "interlaced image" is an image of a frame consisting of two fields separated by the capture time. In the coding and decoding of an interlaced image, it is possible to treat a frame

as a frame itself, as two fields or as a frame structure or a block-by-block field structure within the frame.

The image to be encoded by intra-image prediction without reference to any image will be called image I. The image to be encoded by inter-image prediction with reference to only one image will be called image P. The image to be encoded by prediction. interimage with reference to two images at the same time will be called image B. It is possible that an image B refers to two images that can be arbitrarily combined from progressive and regressive images in order of presentation. Reference images can be determined for each block that serves as the basic encoding / decoding unit. Said reference images can be differentiated by calling a reference image that will be described earlier in an encoded bit stream "first reference image", and referring to a reference image that will be described later in the reference flow "second reference image" bits It should be noted that a condition for encoding and decoding these types of images is that the images used as a reference must already be encoded and decoded.

P and B images are encoded by interimage prediction with motion compensation. Interimage prediction coding with motion compensation is a coding procedure in which motion compensation is used in interimage prediction coding. Unlike a procedure for prediction based simply on the pixel values ​​of a reference image, motion estimation is a technique capable of increasing the accuracy of the prediction, as well as reducing the amount of data by estimating the amount of movement (hereinafter referred to as "motion vector") of each part of an image, and also making a prediction in which said amount of movement is taken into consideration. For example, it is possible to reduce the amount of data through motion compensation by estimating the motion vectors of the image to be currently encoded and then encoding the prediction residuals between the prediction values ​​obtained by changing only the amount of the respective motion vectors and the image to be currently encoded. In this technique, motion vectors are also recorded or transmitted in encoded form, since the information of the motion vectors is needed at the time of decoding.

The motion vectors of each macroblock are estimated. More particularly, a macroblock of the image that is currently to be encoded will be set, to estimate the motion vectors by calculating the position of the most similar reference block of said macroblock set within the search area of ​​a reference image.

Figure 1 is a diagram illustrating an example of data structure of a bit stream. As depicted in Figure 1, the bitstream has a hierarchical structure as indicated below. The bit stream (stream) is made up of more than one group of images (GOP). Using GOPs as basic coding units, it is possible to edit a moving image, as well as to perform random access. Each GOP is composed of a plurality of images, which can be images I, images P and images B. Each image is also composed of a plurality of segments. Each segment, which is a band-shaped area of ​​each image, is composed of a plurality of macroblocks. On the other hand, each flow, GOP, image and segment comprises a synchronization signal (sync) to indicate the end point of each unit and a heading (header) that is common data to each of said units.

It should be noted that, when the data is not transmitted in a bit stream that is a sequence of continuous streams, but in a packet or similar that is a fragmentary unit, the header and the data part, which is the remaining part, They can be transmitted separately. In this case, the header and the data part will not be integrated in the same bit stream represented in Figure 1. In the case of a packet, however, although the header and the data part are not transmitted in sequence, the heading corresponding to the data part is simply transmitted in another packet. Therefore, although the header and the data part are not integrated in the same bit stream, the concept of coded bit stream described with reference to Figure 1 is also applicable to packets.

In general terms, people's sense of sight is more sensitive to low frequency components than to high frequency components. In addition, since the energy of the low frequency components of an image signal is greater than that of the high frequency components, the coding of images is performed in order of least to highest frequency of the components. As a result of the above, the number of bits necessary to encode the low frequency components is higher than that necessary for the high frequency components.

Taking into account the previous points, the available coding procedures use, for the high frequency components, quantification steps longer than those of the low frequency components when the transformation coefficients, which are obtained by orthogonal transformation, are quantified. the respective frequencies. This technique has allowed conventional coding procedures to achieve a large increase in the compression rate with a small loss of image quality from the viewer's point of view.

Meanwhile, since the sizes of the quantization stages of the high frequency components with respect to the low frequency components depend on the image signal, a technique has been conventionally used to change the sizes of the quantization stages for the respective frequency components of each image. A quantization matrix is ​​used to obtain the quantification steps of the respective frequency components. Figure 2 represents an example of a quantification matrix. In this drawing, the component in the upper left corner is a direct current component, while the components on the right are horizontal high frequency components and the downstream components are vertical high frequency components. The quantification matrix in Figure 2 also indicates that longer quantification stages are applied to the higher values. Usually, it is possible to use different quantization matrices for each image, describing the matrix to be used in the heading of each image. Therefore, even if the same quantization matrix is ​​used for all images, that matrix is ​​described in the heading of each image and transmitted individually.

Meanwhile, the current MPEG-4 AVC standard does not include the quantification matrix as in MPEG-2 and MPEG-4 standards. This makes it difficult to achieve optimal subjective quality in the current MPEG-4 AVC coding system and other systems that use uniform quantification in all DCT or DCT type coefficients. When such a quantization matrix system is introduced, the current provisions of the MPEG-4 AVC standard or others should be allowed to understand the quantification matrices, for consideration of compatibility with existing standards.

In addition, due to the increase in coding efficiency, the MPEG-4 AVC standard has been able to offer the potential to use in various application domains. Versatility guarantees the use of different sets of quantization matrices for different applications; different sets of quantization matrices for different color channels, etc. Encoders can select different quantization matrices depending on the application or the image to be encoded. For this reason, it is necessary to develop an effective definition and protocol for loading the quantization matrix to facilitate a flexible but effective transmission of the quantization matrix information.

The article entitled "Quantization Tools for High Quality Video" published according to the guidelines of ISO / IEC JTC1 / SC29 / WG11 and ITU-T SG16 Q.6 indicates that it has been proposed to extend the quantification, so that it is compatible with the high video quality, through a weighting matrix and an expanded quantification table. An encoder adaptively uses a weighting matrix and encodes it in a bit stream. The QP range is extended to -8 and new weighting matrices are formed. A weighting matrix is ​​best transmitted out of band and through the parameter sets of the NAL layer. A weighting matrix must be changed to a position higher than the image layer. The extension of the quantification table combined with the quantization matrix is ​​necessary for high quality video coding.

The article entitled "New Quantization Tools" published in the framework of ISO / IEC JTC1 / SC29 / WG11 deals with a similar object and indicates that, in an attempt to normalize the new video coding technology, tests were carried out that demonstrated, with Regarding compatibility with high video quality, that quantification poses several problems.

The article entitled "Adaptive image compression using local pattern information", published in Pattern Recognition Letters 23 (2002), pages 1837 to 1845, describes a new coding technique adapted to the coding of transform coefficients used in compression systems of block-based images. The presence and orientation of the border information in a subblock are used to select different quantization tables and zigzag scanning paths to cover the design of the local image. The measures of edge presence and edge orientation in a subblock are calculated outside their DCT coefficients, and each subblock can be classified into four different border designs. Experimental results show that, compared to JPEG encoding and enhanced HVS-based coding, the new system has significantly increased the compression rate without deteriorating the quality of the reconstructed image.

Exhibition of the invention

The present invention has been conceived taking into account the above circumstances, the objective of the present invention being to offer a method of decoding moving images that is capable of reducing the amount of data to be encoded and increasing the coding efficiency.

The method of decoding moving images according to the present invention is an image decoding process according to claim 1.

According to the above procedure, it is possible to decode a coded continuous stream, in which only the matrix ID is assigned to identify the quantization matrix used for quantification to predetermined units, such as image units, segment, macroblock or the like, while the quantification matrix has been previously transmitted separately.

In the previous procedure, each image is composed of a luma component and two types of chroma components, and in case there is no quantization matrix in the identification for a chroma component type corresponding to the current decoding in the quantization matrices identified based on the identification information, a quantification matrix of another type of chroma components can be identified for use as a quantification matrix.

According to the above procedure, it is possible to decode a coded continuous stream even if there is no quantization matrix for the chroma components.

As it is obvious from the above description, according to the motion image decoding process of the present invention, it is possible to reduce the amount of data to be encoded and offer effective encoding and decoding.

Brief description of the drawings

These and other objectives, advantages and features of the present invention will become apparent from the following description thereof, taken in conjunction with the accompanying drawings illustrating a particular embodiment of the present invention. In the drawings:

Figure 1 is a diagram illustrating an example of data structure of a bit stream;

Figure 2 is a diagram depicting an example quantization matrix;

Figure 3 is a block diagram representing the structure of a motion image coding apparatus that constitutes an embodiment of the motion image coding method according to the present invention;

Figure 4 is a diagram representing the correspondence between the sequence parameter sets and the image parameter sets and the images;

Figure 5 is a diagram representing a part of a structure of a set of sequence parameters;

Figure 6 is a diagram representing a part of a structure of a set of image parameters;

Figure 7 is a diagram representing an example of description of the quantization matrices of a set of parameters;

Figure 8 is a flow chart representing the operations to assign an array ID;

Fig. 9 is a block diagram showing the structure of a motion picture decoding apparatus that constitutes an embodiment of the motion picture decoding method according to the present invention;

Figure 10 is a flow chart depicting the operations to identify a quantification matrix;

Figure 11 is a flow chart depicting the operations to identify a quantification matrix for chroma components;

Figure 12 is a diagram representing the correspondence between quantization matrices transmitted as separate data and quantization matrices for sequences;

Figures 13A to 13C are diagrams illustrating recording means in which a program is stored for carrying out, by means of a computer system, the method of decoding moving images according to the previous embodiment, and particularly Figure 13A It is a diagram illustrating an example of a physical format of a floppy disk as the main body of recording media Figure 13B is the complete aspect of the floppy disk seen from the front of it, a cross-sectional view thereof and the floppy disk itself, and Figure 13C is a diagram illustrating a structure for recording on the floppy disk and reproduce from this the previous program;

Figure 14 is a block diagram showing a general configuration of a content delivery system that constitutes an embodiment of a content distribution service;

Figure 15 is a diagram depicting an example cell phone;

Figure 16 is a block diagram representing the internal structure of the cell phone and

Figure 17 is a diagram representing the overall configuration of a digital broadcasting system.

Best way to practice the invention

The embodiments of the present invention are described with reference to the diagrams.

[First embodiment]

Figure 3 is a block diagram depicting the structure of a moving image coding apparatus.

An image coding apparatus 1 is an apparatus for performing compression coding of an input image signal Vin and generating a continuous stream encoded Str that has been encoded as a bit stream applying variable length coding and the like. As shown in FIG. 3, said image coding apparatus 3 consists of a movement estimation unit 101, a movement compensation unit 102, a subtraction unit 103, an orthogonal transformation unit 104, a quantification unit 105, an inverse quantization unit 106, an inverse orthogonal transformation unit 107, an addition unit 108, an image memory 109, a switch 110, a variable length coding unit 111 and a quantization matrix reserve unit 112.

The Vin image signal is transmitted to the subtraction unit 103 and the motion estimation unit 101. The subtraction unit 103 calculates residual pixel values ​​between each image of the input image signal Vin and each predictive image, and transmits the calculated pixel values ​​to the orthogonal transformation unit 104. The orthogonal transformation unit 104 transforms the residual pixel values ​​into frequency coefficients and transmits them to the quantization unit 105. The quantization unit 105 quantifies the received frequency coefficients through the received WM quantization matrix and transmits the resulting quantized values Qcoef to the variable length encoding unit 111.

The inverse quantization unit 106 performs the inverse quantification of the quantified values ​​Qcoef by the received quantization matrix WM, to convert them into the frequency coefficients, and transmits them to the inverse orthogonal transformation unit 107. The inverse orthogonal transformation unit 107 performs the inverse frequency transformation of the frequency coefficients to transform them into residual pixel values, and transmits them to the addition unit 108. The addition unit 108 adds the residual pixel values ​​and each Predictive image generated by the motion estimation unit 102, to form a decoded image. The switch 110 is deactivated when it is indicated that said decoded image should be stored and that said decoded image should be stored in the image memory 109.

Meanwhile, the motion estimation unit 101, which receives the Vin image signal from each macroblock, detects the image area closest to an image signal in said Vin image signal received within a decoded image stored in memory of images 109, and determines one or more MV motion vectors that indicate the position of said area. Motion vectors are estimated for each block, which gets even more a macroblock. Once this is done, it is possible to use more than one image as reference images. In this case, since a plurality of images can be used as reference images, identification numbers (Index reference indexes) are needed to identify the respective reference images of each block. Using the Index reference indexes, it is possible to identify each reference image by associating each image stored in the image memory 109 with the designated image number to each of said images.

The motion compensation unit 102 selects, as a predictive image, the most suitable image area of ​​the decoded images stored in the image memory 109, using the motion vectors detected in the previous processing and the Index indexes.

The quantization matrix reserve unit 112 reserves the WM quantization matrix that has already been transmitted as part of a set of parameters and the matrix ID that identifies this WM quantization matrix in the manner in which one is associated with the other.

The variable length coding unit 111 obtains, from the reserve unit of quantization matrices 112, the matrix ID corresponding to the quantization matrix WM used for quantification. The variable length coding unit 111 also performs the variable length coding of the quantification values ​​Qcoef, the matrix IDs, the Index reference indexes, the Ptype image types and the MV motion vectors, in order to obtain a continuous stream encoded Str.

Figure 4 is a diagram representing the correspondence between the sequence parameter sets and the image parameter sets and the images. Figure 5 is a diagram representing a part of a

structure of a set of sequence parameters, and Figure 6 is a diagram representing a part of the structure of a set of image parameters. Although an image is composed of segments, all segments included in the same image have identifiers that indicate the same set of image parameters.

In the MPEG-4 AVC standard, the concept of heading does not exist, and the common data is placed at the top of a sequence below the designation of the parameter set. There are two types of parameter sets, the PPS image parameter sets that consist of the data corresponding to the header of each image, and the SPS sequence parameter sets corresponding to the header of a GOP or a sequence in MPEG-2. A set of SPS sequence parameters comprises the number of images that are available as reference images, the size of the images, etc., while a set of PPS image parameters comprises a type of variable length encoding (which changes between Huffman coding and arithmetic coding), the predefined values ​​of the quantization matrices, the number of reference images, etc.

An identifier is assigned to a set of SPS sequence parameters and, to allow identification of the sequence to which an image belongs, this identifier is indicated in a set of PPS image parameters. An identifier is also assigned to a set of PPS image parameters and, to allow identification of the set of PPS image parameters to be used, this identifier is indicated in a segment.

For example, in the example depicted in Figure 4, image # 1 comprises the identifier (PPS = 1) of a set of PPS image parameters to be indicated in a segment comprised in image # 1. The set of PPS image parameters # 1 comprises the identifier (SPS = 1) of a set of sequence parameters to which reference is to be made.

In addition, the set of SPS sequence parameters and the set of PPS image parameters, respectively, comprise signaling devices 501 and 601 indicating whether or not quantization matrices are transmitted, as depicted in Figures 5 and 6 and, in if quantification matrices are to be transmitted, a respective description of quantification matrices 502 and 602 is added.

The quantization matrix can be adaptively changed with respect to the unit of quantification (for example, 4 horizontal pixels x 4 vertical pixels and 8 horizontal pixels x 8 vertical pixels).

Figure 7 is a diagram depicting an example of description of the quantization matrices of a set of parameters.

Since a Vin image signal consists of luma components and two types of chroma components, it is possible to use different quantization matrices for luma components and the two types of chroma components separately when quantifying. It is also possible to use different quantization matrices for intraimage coding and interimage coding separately.

Accordingly, for example, as shown in Figure 7, it is possible to describe the quantization matrices for a quantification unit, the luma components and two types of chroma components and the intraimage coding and interimage coding, respectively.

The following describes the operations to assign the matrix IDs in the structured motion image coding apparatus in the manner described. Figure 8 is a flow chart representing the operations to assign an array ID.

The variable length coding unit 111 obtains a quantization matrix WM used for quantification (step S101). Next, the variable length coding unit 111 determines whether or not the quantization matrix WM obtained is in the quantization matrix reserve unit 112 (step S102). Then, in case the quantization matrix obtained WM is found in the quantization matrix reserve unit 112 ("Yes" in step S102), the variable length coding unit 111 obtains the matrix ID corresponding to the Quantification matrix obtained WM in the quantization matrix reserve unit 112 (step S103). Then, the variable length encoding unit 111 assigns the array ID obtained to predetermined units (for example, to each image, segment or macroblock) (step S104).

On the other hand, in case the obtained WM quantization matrix is ​​not found in the quantization matrix reserve unit 112 ("No" in step S102), the quantization matrix reserve unit 112 generates the ID of matrix for this WM quantization matrix (step S105). Next, the quantization matrix reserve unit 112 reserves this quantification matrix WM and the matrix ID in the manner in which they are associated with each other (step S106). The variable length coding unit 111 assigns the matrix ID obtained to predetermined units (for example, to each image, segment or macroblock) (step S107). The variable length coding unit 111 describes the generated matrix ID and the quantization matrix

WM in the parameter set (step S108). It should be noted that the set of parameters in which the matrix ID and the quantification matrix WM are described is transmitted, in a continuous stream encoded Str, before the predetermined units (ie, the data encoded quantified by the quantization matrix WM) to which said matrix ID is assigned.

As described above, since the WM quantization matrices are described in a set of parameters and transmitted, while only the matrix ID identifying the WM quantization matrix used is assigned to the predetermined units (for example, to each image, segment or macroblock), it is not necessary to describe the WM quantization matrix used in each predetermined unit. Therefore, it is possible to reduce the amount of data to be encoded and offer efficient coding.

It should be noted that it is possible to update a WM quantization matrix transmitted in a set of SPS sequence parameters and transmit the updated matrix (with the same matrix ID) in a set of PPS image parameters. In this case, the updated quantization matrix WM is used only when the PPS image parameter set is indicated.

It is also possible to integrate into a continuous coded stream a signaling device indicating the change between the predefined quantization matrix WM and the quantization matrix WM identified by a matrix ID. In this case, the predefined quantization matrix WM is replaced by the quantization matrix WM identified by the matrix ID according to the flag.

Fig. 9 is a block diagram showing the structure of a motion picture decoding apparatus that constitutes an embodiment of the motion picture decoding method according to the present invention.

The motion picture decoding apparatus 2 is an apparatus that decodes a coded continuous stream obtained by encoding by the motion picture coding apparatus 1 described above, and comprises a variable length decoding unit 201, a unit of reserve of quantization matrices 202, an image memory 203, a motion compensation unit 204, a reverse quantization unit 205, an inverse orthogonal transformation unit 206 and an addition unit 207.

The variable length decoding unit 201 decodes the encoded continuous stream Str and generates quantified Qcoef values, Index reference indexes, Ptype image types and MV motion vectors. The variable length decoding unit 201 also decodes the encoded continuous stream, identifies a quantization matrix WM based on the extracted matrix ID and facilitates the identified quantization matrix WM.

The quantization matrix reserve unit 202 associates the WM quantization matrix that has already been transmitted in a set of parameters with the matrix ID that identifies this WM quantization matrix and reserves them.

The quantified values ​​Qcoef, the Index reference indexes and the MV motion vectors are entered in the image memory 203, the motion compensation unit 204 and the inverse quantization unit 205, and then decoding processing is applied to these . The decoding operations are the same as those of the motion picture coding apparatus 1 shown in Figure 3.

Next, the operations for identifying a quantization matrix in the decoding apparatus of structured motion images as described above are described. Figure 10 is a flow chart representing the operations to identify a quantification matrix.

The variable length decoding unit 201 decodes a continuous stream encoded Str and extracts an array ID assigned to predetermined units (step S201). Next, the variable length decoding unit 201 identifies a quantization matrix WM of the quantization matrices reserved in the quantization matrix reserve unit 202, based on the extracted matrix ID (step S202). Next, the variable length decoding unit 201 transmits the identified quantization matrix WM to the inverse quantization unit 205 (step S203).

As described above, although WM quantization matrices are described in a set of parameters and transmitted, in predetermined units (for example, in each image, segment or macroblock), it is possible to decode a coded continuous stream to which only You have assigned the matrix ID that identifies the WM quantization matrix used.

It should be noted that, although in the present embodiment the WM quantification matrices are described in a set of parameters and transmitted, the present invention is not limited to this case. For example, quantization matrices can be previously transmitted separately from a coded continuous stream.

Since a Vin image signal consists of luma components and two types of chroma components as described above, it is possible to use different quantization matrices separately for luma components and the two types of chroma components when quantifying. It is also possible to use a uniform quantization matrix for all components.

Next, the operations to identify the quantification matrices to be used for chroma components are described. Figure 11 is a flow chart depicting the operations to identify the quantization matrices to be used for the chroma components.

The variable length decoding unit 201 finds out whether or not there is a quantization matrix for the type of chroma components corresponding to the current decoding between the identified WM quantization matrices as mentioned above (step S301). For example, if the quantified value Qcoef to be decoded is a first chroma component, it is determined whether or not there is a quantization matrix for the first chroma components. If the quantified value Qcoef to be decoded is a second chroma component, it is determined whether or not there is a quantization matrix for the second chroma components. Then, if there is a quantization matrix for the corresponding chroma component type ("Yes" in step S301), the corresponding chroma quantization matrix is ​​transmitted to the inverse quantization unit 205 for use (step S302).

On the other hand, if said corresponding chroma quantification matrix does not exist ("No", in step S301), the variable length decoding unit 201 determines whether or not there is a quantization matrix for another type of chroma components (step S303 ). For example, if the quantified value Qcoef to be decoded is a first chroma component, it is determined whether or not there is a quantization matrix for the second chroma components. If the quantified Qcoef value to be decoded is a second chroma component, it is determined whether or not there is a quantization matrix for the first chroma components. Then, if there is a corresponding quantization matrix for another type of chroma components ("Yes" in step S303), the quantization matrix for another type of chroma components is transmitted to the inverse quantization unit 205 for use by it ( step S304). On the other hand, if there is no quantization matrix for any other chroma component ("No" in step S303), the quantization matrix for the luma components is transmitted to the inverse quantization unit 205 for use by it ( step S305).

It follows that it is possible to decode a coded continuous stream even if there is no quantization matrix for chroma components.

Industrial applicability

As described above, the motion image coding process and the motion image decoding method according to the present invention are useful as methods for encoding the images that make up a moving image in order to generate a continuous coded stream. and to decode the coded continuous stream generated, in devices such as a cell phone, a DVD device and a personal computer.