Improved block transform and quantization for image and video coding

The conversion device and application quantization improved the image and video encoding using are the same flat of application on September 18, 2002, the application of 02143205. 8, the back of the conversion device and quantization improved an image and video encoding for of the adjacent application of patent of invention. Domain present invention claims a coding and decoding of image and video signal, generally special relates to a method for encoding and decoding image video signal conversion device and inverse conversion improved, and quantization and adapting to the quantization method.

Invention independent background cosine conversion and image and video image (sequence code) is conversion device, for example JPEG and MPEG. Other end is; the luminescence image quality, DCT a random signal (e.g., corresponding to image data signals) separated for gate low-frequency component and secondary frequency unit, to the conveniently for compression.

For example, H.26L is for 4*4 pixel block with the frame images and/or the video encoding method, which uses 8X8 pixel block MPEG and JPEG are different; AN. 26L by using estimation and motion compensation a frames (. i.e, the edge code) evening divination, the blocking predicting system according further - comprising is provided with a frame (. i.e, a frame code), wherein with the 4*4 pixel block to obtain relatively the high compression. The general, so as to achieve the inter-layer prediction, the &261 encoder claim tell the corresponding decoders 11.261, which beforehand pixel blocks in a frame served and is a predicting base, and a or 6 possibilities of the predictors type (1) to determine the novel block pixel block (pixel from the front recognized). The fixed bring and prediction error; the error further providing decoder, adjustment novel block. The prediction error information converting (DCT) is code by block, a coding result is, decoder for counting error information, comprising adjusting the predicting through the inverse conversion once more block.

The image and video encoding or and decoding circuit massively the processing capacity. Which is a known, processing capacity of the encoder with flowing is greater than multiple converting data into the is processing capacity, wherein the decoder usually providing a user equipment, and, image and video decoding a occur the television top box, a personal digital assisting, large and computer and in multi-cell improved cellular telephone.

Therefore, considered and image and video coding and decoding method, and simplicity for decoding is very gate, and maintain the method for coding flowing connected with the computer is more complicated. Any measuring of simplified decoding is a circuit, which has, processing of encoding and decoding it has a adverse impact on the amounts of assigned with prior a fundamental realizing compression horizontal terminal and image quality and decoder circuit.

invention content

The short, the invention claims a method for improving, system and block conversion, so the encoder and decoder and capable of simplifying considerably and video the computing complexity to the. And the influence of simultaneously, a compressing and image the/video made and resource is a lower neglect.

Transformer, claims with the novel integer four orthogonal conversion device and corresponding inverse conversion to independent cosine conversion, and methods of significantly of the encoder and decoder computing complexity computational exchange. The fact, in one embodiment, wherein the encoder and decoder; each coefficient's conversion operations value of the secondary reduced to 4 in addition to 16 and 1 shift (; the NOTCHES. cover 26L claim stipulated is 4 times additions and 3 multiplications in operations 32). Wherein during code quantization (conversion and adjusting and decoding the substrates quantization to turn) through a shaft according to one of forming 3 are arranged at the) is the conversion doubling coefficients in the compensation the parameter value (is a conversion and inverse conversion processing locating function of doubling according completes); therefore the invention can converting pixel information correctly (for example, error correction) data.

Moreover, the decoder to the quantization using time 1 to 16 multiplications during and decoding, wherein pass through the quantization the locations to go to the quantization, wherein the conversion ratio, 32 multiply operations for polymeric (for example H.26L lower). Multiple device, a universal image and video, the invention the advantage and H.26L of data - processing compare are status, special and decoder; and there is one consumption of quality compression and/or, and neglect.

The invention claims a device for decoding a video or the pictorial information method, wherein the method comprises: Method for quantization coefficient; The adapting to the telemetering signal to the photoelectric conversion quantization ratio of quantization coefficient; And using inverse conversion processing in the propping the quantization the conversion ratio, the inverse conversion computational result of the inverse conversion processing comprising generating a given of the following: equalityu =A interface IS one A-c, 'y the (B >> 1) z the (D( 1) =u interface; z.

b' =V y interface, ', c; =V-y

d' =u z; the A, B, C and D representative flowing of the inverse conversion conductor, u, v, and y represents z the subsidiary variable, and a' b', and c and d'代表 the inverse switching value.

The invention claims a device for coding to the video or the pictorial information method, wherein the method comprises: Application conversion processing message to a conversion ratio, a conversion processing comprising generating a conversion Computation-u result of the following equality a given the one +, d of the v of ethylene propylene interface,' y the ethylene propylene wherein; z, a-d A u the operating v; C the u - v,' AND the y interface (z<p>v the; A-c

the y (B >> 1) is d;

the z (D >> 1) interface. ; THE

a' the u Z' interface.

b' the v y interface, a c; the v of y.

d; the u-z, the A, B, C and D representative flowing of the inverse conversion conductor, u, v, and y represents z the subsidiary variable, and a and b, a c and d'代表 the inverse switching value.

From the following attached with figure's detailed description, and methods and merits are status: The auxiliary digital instructional diagram is 1 to show comprises the invention block diagram of computer display system,Digital is 2 to show to act according to the invention claims a surface, and coding video or the image data pixel the video according Jing For coding, at least penetrates through the video decoding of code for pixel information the block diagram of part, comprising a conversion, a quantization; and quantization and inverse conversion.

Digital 3 cooling to act according to the improved conversion matrix butterfly-shaped structure of this invention claims a surface unit. Digital 4 cooling to act according to the improved inverse conversion matrix butterfly-shaped structure of this invention claims a surface unit.

detailed description of illustrated embodiments

Digital 1 cooling relate to implement the invention to loudspeakers the work environment 120 examples, especially and/or the decoder of video data to image. The work environment 120 loudspeakers operating environment an example, to the invention application mode or a filler claims according to a. Suitable for the well-known computing system, environment and/or a configuration the present invention includes (, it does not to limit): The personal computer and computer server, and hand-held laptop device and multiprocessor system, user electronic installation-based and network of PC machine, the minicomputer and a micro computer processing system, a code, comprising any one or system of distributed computing environment equal. For example, image and/or the code of video image data very possibly unusually to 1 and than a handheld nowadays personal computer stronger processing capacity computer and realizes; which is averaging it is a on the code on the display device, or and decoding on stronger processing ability's device.

The indication range of the invention the common computer and for fabricating, to the program module, a executing is one or more computer or other device. The general, program module comprises a bearing (task or the specific abstraction data stream device, program, object, constituent, data structure equal. Generally program module filler performing a combined or a distribution according to the examples requirements.

Calculating device 120 and at least a part of the computer is connected to the reading media. The media for calculating device 120 and deposits and withdraw the function adding the media be any is a straight. For example, which is not restricted, function adding the media is connected to a computer storage medium and media communication. The storage medium comprises a volatilizing and non-volatile of any method or technology the filling realizing data, removable and intransportable media, to the function adding the instructions, the framework of data, a program module or data. The storage medium includes, which is not restricted, RAM, ROM, EEPROM and quick of flash memory or technology, CD-R0M and digital universal plate (DVD) or the memory, floppy disk, magnetic disc and magnetic-disc memory or magnistorized element, or can be used to save need to the information, and is any other media for calculating device 120 and deposits and withdraw. The communication media generally comprising a computer is connected to the reading instructions, the framework of data, a program module or to the holding or transmission mechanism comprises a data signals of data, and a any message routing media. Terminology is in a shape of data signals of adjusting a wherein the group 1 characteristic or more characteristic or according to of the signal the manner of information coding to the signal of transformer. For example, without limit, communication media comprising a media for example: Wireline network or directly wired connection, and wireless media for example: An audio frequency and RF, an or the wireless medias. To any one of claim, comprising a function adding the media in range.

Digital 1 cooling a such calculating handheld device 120 and function unit, a processor, 122 and 124, 126 monitoring and keyboard 128 )which can be leads or virtual keyboard). Memory 124 — comprising volatile memory (for example RAM) and nonvolatile memory (for example R0M and PCMCIA card equal) are.

The operating system 130) and memory 124, 122) executing the processor, Microsoft to Corporation Window® the CE operating system or operating system.

One or more application 132 load and memory 124, 130 movements the operating system. Application of a e-mail program, scheduler program and PIM (personal information management) is character, image processing method and electronic program of Internet and browser is equal. The handheld personal computer 120 - comprising a load memory 124 informing performing programs 134, wherein the program 122) executing the processor. Informing program performing processing 134 such as informing claim from application.

Handheld personal computer 120 ^^ a power supply is one or more batteries are implemented as 136. Power supply 136 - comprising an external power supply, battery overload (override) or charging to the local, connected to one of AC adaptor or power supply.

Digital 1 the display is personal computer for fixing 120 on the external informing mechanism of length is 3: One or more light-emitting diode 140 and 1 acoustic frequency generator 144. The part of the conductive couplings to power supply 136, wherein, when the peripheral activate, and maintain the handheld personal computer processor 122 and constituents to save the battery power and down; the position of the part located in informing the mechanism of assigned thereof. LED best 140) stops infinite maintaining opening; and action is a user. The warning and an acoustic frequency generator 144 consume the handheld personal computer battery power is too now multiple now, when work and dormant or the activation a timing limit-current; the protective layer.

The coding and decoding on the background, independent cosine conversion using linear polymeric type X-Hx is N vector X mappings a new conversion ratio vector X, AN k ohm and wire element is n ohm bar is stipulated of the differential: aAnd k^), 1,…, N-1, further n-O, 1,… N-1, when ends of > l, c. Are further V^ c one l-shaped. The DCT substrate is orthogonal, therefore, wherein reversion equal to the transposition, wherein x two H-1 X=HTX.

The DCT A disadvantage of the is H (Ends, N) is one of irrational, therefore integer data-in x (n) in map for unreasonable conversion ratio X (K). Therefore, when using digital computer, which directly and inverse conversion a cascade calculating method, data output on non-closely coincide the data-in. The other testing; and calculates of X^Hx, and u-round (HT X), to the n, u (n divination x (n) is a vacuum. The is provided with a coefficient of reducer, a Y-SHAPED, a invention x-yHx, further u-round (aHTX); and a Y-SHAPED and enough one large manner, a almost the n, u (n) is Gx (n), (of the turn; the it for ensuring consistency between datum of the correcting result.

A any video encoder motion compensation, for example, the data frame according decoded service and is used to generate a current the frame for information reference information. Therefore, and a part of code, the encoder generating wherein for decoding the frame, wherein the encoder need to equal the inverse conversion. And application type u=round (ocHTx); and floating different joint format and adjusting method of different a processor for different result. The plurality of; the decoding data on the encoder and decoding data and decoder fixed with a deviation (solution with different processor), wherein the encoder based on to the frame is made prediction/movement information; the block is composed fewer and fewer frame, such block according and decoder generating, therefore each l novel frame images is composed of multiple and a first.

For data offset method and installed with an integer substrate is similarly rectangular H, capable of removing rounding error. A H lines are quardratures and same norm (sum of square); and following a follows to are integers x, a equal u of the turn operation correctly. The other, testing, wherein the direct contact is calculated based on X^x, wherein the inverse conversion based on u=HTX is calculated, obtaining result of u=Gx, A are integers, equal to AN any line of square norms.

The DCT production integer mounting value's a method of use universal formula: (Q ends, n) =round (ccH k (, n))And oc is a conversion parameter.

The H.26L compressing video transmission, the is composed of a 4*4 pixel block; therefore the 8.261^ 0 ): Ding the N type two 4. DCT matrix's result more than one

<img id="idf0001" file="CN100463522CD00101.tif" img-content="drawing" img-format="tif" />

and C V^cos(u/8) and S V^sin (high-rising /8).

With a version NOTCHES. in 26L the conversion substrate is of oc =26 obtaining; the value computer is:

<img id="idf0002" file="CN100463522CD00102.tif" img-content="drawing" img-format="tif" />

A variable parameter value 26, the left. The wire and bar which are quardratures (. i.e, any two rows of inner product and zero), the line and groove is 1 norm 676. The fact, a a &lt; 50:00, only and c^2 or a-26 can provide with a equal norm line parallel matrix. Compression effect of the cc-2 for obtaining is not rod, and more a value is not satisfying, wherein calculates the direct conversion result X are FORMED. wherein x, a computing complexity and lower fixed wing (for example, word) is. Therefore the &amp;=26 of every 261^ is now.

Inverse conversion is FORMED x'二. X-RAY stipulated of therefore available integer operation calculates. To the stipulation X, 2676 ' x, Gen trigram P, the data of restructuring x'等于 the primary data x is increased 676) is connected to enlarge (comprising a integer is FORMED. any line of norms). To the invention surface, claims a uses an integer terminals of the conversion device array; the mounting values and quardratures, with different value; and when coding and decoding significantly reducing complexity computing. Is special, and a-2.5, a production the column matrix:

<img id="idf0003" file="CN100463522CD00111.tif" img-content="drawing" img-format="tif" />

Warning: The Qn line of the middle orthogonal, and methods norm is different line (0 and line of norms 2 and 4; Of 1 line and line of 3 norms are 10), a as follows clearly, which is increasing the complexity computing method for quantify is an and adapting to telemetering signal processing.

Loop comprises a gasket digital 2, wherein the display of the block conversion encoder 200 and decoder (222, 220 and connected with the partial medium and/or the middle device. Which is a known, the encoder 200) no need by the output of the decoders 222, wherein the wire of, digital section 2 to 220 of virtual represent to media or a element, to network cable, wireless transmission media, telephone wire; and router, and/or methods virtual any combinations.

The general, the encoder 200 transform 204 the conversion according to each NXN input pixel data 202 is separately 2 maintaining-operating work station. Is special, a flexible is 206 to converting block group of a pixel data by a conversion circuit; and wire of transformations' result digital (2) is not shown to process 208 to converting results from copper by a conversion special 210). The other testing, 2, DCT maintaining-operating work station substantially and a maintaining-operating work station 1 DCT to the pixel block wire, comprising, and is an expert the flat pixel blocks on generating end is 1 DCT maintaining-operating work station is 1 DCT maintaining-operating work station. Line and flat transformations is a on a are connected, and result for obtaining are the same.

The H.26L; the conversion substrate according stipulated of the following formula, a equal conversion: ratio

<img id="idf0004" file="CN100463522CD00112.tif" img-content="drawing" img-format="tif" />

, Two abed) first represents the rod pixel value of the 4*4 block; the is an expert the conversion, a representing the line of transformations, data of the pixel value. And a reality; the formulas capable of simplifying, the DCT substrate with recursive structure, capable of reducing doubling and region of the value of secondary request. The, the substrate needs at least 4 additions and 3 multiplications equal the picture element the conversion ratios. Additionally, wherein every pixel possibly is a symbol 9 values shape; the pawnshop and factor (gain is equal to norm 676), doubling calculates the other operations 32, wherein the independent 32 operations claim spend and multiple 16 operating same time. When the consideration factors are is generally status during code, and a NOTCHES. the matrix values of 26L stipulated for with extra operations and 32 operations during decoding, therefore, the overhead messages is very high.

Is opposite with the H.26L stipulation's matrix, using same universal type method for portrayal and, wherein the diaphragm component conversion substrate is:

<img id="idf0005" file="CN100463522CD00121.tif" img-content="drawing" img-format="tif" />

Providing comprising the following formula is equal the conversion: ratio

<img id="idf0006" file="CN100463522CD00122.tif" img-content="drawing" img-format="tif" />

The type/matrix of optimize part end of the through to form the value of the conversion for simply, wherein the phase-discrimination factor substantially to process for quantization table, wherein substitutes disposed on each doubling the switching table. Finally; and by the only doubling is multiplied of power 2, wherein the operation the processor completes through the simple transfer left operation, is not the true doubling. A u, v, and y and z auxiliary parameter, butterfly-shaped structure of an electronic watch working state and file page 12&lt;/formula&gt; as a digital 3 (line instructions and numbers 2 o'clock, and multiplex digital multiple (l, 2 and 2 pointed of the phase-discrimination factor), search the conversion ratios the fact without to need to each distribution pixel end is 4 times region and locating 1, wherein compares to the power of H.26L stipulation, capable of providing remarkable improving performance. Wherein the night-watch made of low, the conversion operations of 16 in operations alarm (9) is pixel) data. Aluminium; and conversion array FORMED with a stipulated. , A maintaining-operating work station 2 2-D holes; the focusing signal amplifying factor is 36. From, a maintaining-operating work station 2 voltage; the output rate of claim from 6 to of the input. Therefore, 9) inputs; the output power of 15, dynamiranges is 2 maintaining-operating work station direct contact (16) operations calculates. Moreover, as follows, wherein the operation value of times of reduced; and each time is needed the reinforced performance of 16 operations branches during use the inverse conversion terminal and decoding, and reappearing performance enhance the once; the large enhanced and decoding performance.

Once certain of the conversion ratio 210, 210 to the device using quantifies solution for producing recent integers by regarding the value conversion and adjustable. The quantization process is 212 shown of digital 2, wherein the FPGA for selecting from the quantization parameter 214 quantization values, conversion ratio 210 conversion for quantization coefficient 216. Is as follows, wherein which is an l-shaped norms the improved matrix, which is norms 3, 3 type number and Q0 Q1, Q2 and a values, which are fixed by for quantization conversion the ratio of the block.

The conversion and adjustable quantization base is as follows, wherein the flowing equal is 32: precisionAN two two KXA (QP, r) +f modules( 20) is L-SHAPED the quantization coefficient; K is the conversion ratios, A (QP, R) is a quantization parameter QP and is a coefficient of reducer, which form a r identification handle (QO Q1, Q2 or), fX the two 0-0. 5) X22° in a range (fX provided with a K same mark), further &gt; &gt;20 is the cross 20 core (1,048,576), is made of proportionate thereof. The quantization method therefore introduces the error. Warning, although applied operations 32 to quantify; the is only the lower code, and decoding it is provided with importantly (adapting to quantization is allocated to 16, precision is as follows).

Positioning is from 3 tensor the Q0, Q1, Q3 is l-shaped for acting according to the block k at the position adjusting 3) norms: A made from one of KO, O), (O, l), (1,0); (1,1) DEG, which is two PIN (is QO table); A made from the following position {(0,2); (0.3); (1,2); (1,3); (2,0), (2, 1), (3,0); (3,1) DEG the; and r-l; and Q1 table; A made from one of K2,2), (2,3); (3,2); (3,3) DEG the; and r=2; and Q2 table.

The cross 20, fits the quantization result OF the is a integers 16 in. The same displaying the invention, according to the following form; the locations of The reducer (QP, R) is fixed by quantization parameter QP and coefficient of group is:

A (QP-0..31, r=0) is the {104858, 93418, 83226, 74146, 6S056, 58849, 52429, 46709, 41613, 37073, 33028, 29425, 2S214, 23354, 2080S, 1853S, 16514, 14712, 13107, 11S77, 10403, 9268, 8257, 735S, 6554, 5839, 5202, 4634, 4129, 3678, 3277, 2919 DEG, A '(QP=0..31, r=l) is the {66318, 59082, 52636, 46894, 41778, 37220, 33159, 29541, 26318, 23447, 20889, 18610, 1S579, 14771, 13159, 11723, 10444, 9305, 8290, 7385, S580, 5862, 5222, 4S52, 4145, 3S93, 3290, 2931, 2S11, 2326, 2072, 184S DEG;

A (QP=0..31, ir=2) is the {41943, 373S7, 33290, 29658, 26422, 23540, 20972, 18684, 16645, 14829, 13211, 11770, 10486, 9342, 8323, 7415, 6606, 5885, 5243, 4671, 41S1, 3707, 3303, 2942, 2621, 2335, 2081, -1854, 1651, 1471, 1311, 1168 DEG.

The special value of the design is used for the H.26L meets the design method standard, a quantization of medium every claim 6 increasings when the locating table, doubles. The video frequencies or a position image coding applications, claim consider quality capacity and lower, designs the quantization table. Warning, a latch the table quantization of the encoder, the lower exceptional to space (3), and is further not is 1; the extra amount of space is only bytes 64, therefore is not worthy of mentioning.

A map-read 2, wherein quantization process; the quantization tube with 216 to the entropy encoder 218, wherein when reducing grouping code shape and lower. The entropy (coder decoder and) is everybody knew a very of therefore here of longer tired of the front wherein the invention operation, a substantially with nothing is made of the entropy coding, further comprises an operation, a substantially further provided nothing is made of the entropy decoding.

The consideration when the conversion device of decodes 222, wherein sometimes springs how substrate-processing, entropy coding's output feeding position to the entropy decoder 224 and input of. The general, wherein a entropy decoder known, block telemetering signal regenerating coefficient of therefore the other in operations, and fully invention entropy decoder 224 pairs of 226 assignments. The warning, generally the coding and decoding it harm, namely feeding to the entropy encoder 218 quantization 216 double-face 226) is consistent a quantization coefficient of entropy decoder 224 is a.

The general, block conversion decoder 222 reflective block encoder conversion operations 200, although is as follows, and is consistent with the invention, the invention claims an improved the inverse conversion matrix, can be applied 16 operations in full decoding method, considerably simplified is complexity computing.

The once quantization coefficient 226 224) is out from entropy decoder; each quantization OF coefficient of substrates to the quantization of flexible 228 to for restructuring (adapting to quantization) K' value; and a telemetering signal processing end of the following Formula-k' =LXB (QP, r) are made of reducer AND is fixed in the QP code rate used, or 1 form's l parameters of r is from 3 to form D0, a or Diode D2 are fixed for quantization parameter 230 (adjustment 3) norms. Warning; and related quantization of the r of the arc-shaped quantization of considering from the block, although sent to the r value of the encoder is feasible (that can reduce complexity on decoder calculated); the other at least flowing number of 2 to each tube, so that the number of mark is.

Wherein, when coefficient of reducer (B QP, R) is fixed in the code QP quantization index parameter and at a position of group is provided. For example, according to the lower surface stipulation: relevance(B QP=0..31, r=0) is the {80, 90, 101, 113, 127, 143, 160, 180, 202, 226, 254, 285, 320, 359, 403, 453, 508, 570, 640, 718, 80S, 905, 1016, 1140, 1280, 1437, 1613, 1810, 2032, 2281, 2560, 2874 DEG;

(B QP=0..31, r=l) is the {101, 114, 127, 143, 161, 180, 20.2, 227, 255, 286, 321, 3S1, 405, 454, 510, 572, 643, 721, 810). , 909, 1020, 1145, 1285, 1443, 1619, 1817, 2040, 2290, 2570, 2885, 3239, 3635^B (QP=0-31, r=2 the) (128, 144, 161, 181, 203, 228, 256, 287, 323, 3S2, 40S, 45S, 512, 575, 724, 813, 912, 1024, 1149, 1290,1448, 1S25, 1825, 2048, 2299, 2580, 2896, 3252, 3650, 4095, 459'.

When using code table, the specific value of the design satisfies the H.26L design method standard, namely quantization step's medium flowing is a single indexed each capacity is 6 doubles. To the video frequencies or a position image coding applications, considered of the lower quality the capacity, and design quantization different.

Understandable is although, each values in the conversion values and low enough, a ensuring consistency between datum of the decoder is needed operations 16, further adapting to the quantization type to make the locations to convert into the light value. Are flowing pay warning, wherein each of items 1 table on the corresponding QP index's with one another 2 in different mathematical and related (based on norm), possibly conversion method for manufacturing suitable mathematical adjustment only by 1 to form according to the r of the QP value; and an with 3 to. The 査, wherein looks a table of the motor, and saving 3 to 32 items to be (H) of bytes 2 space. 26L conversion stipulation lower) comprises one saving bytes count of spaces 4 1 and 32 items of filtering circuit is 64 extra bytes, wherein a neglect the modern calculating device, considered special of substrates to the telemetering signal conversion production's ratio fits to 16 of testing, and not to 32 of testing, wherein capable of reducing storage position of need to 2, are once 226 adopts the type of the quantization coefficient andThe table quantifies, a restructuring conversion ratio 232 matrices 4*4. To the conversion ratios 232, wherein with them are respectively processes the flat and good inverse (236, 2 inverse maintaining-operating work station according transformations 238) for processing 234, capable of generating high-energy restructuring pixel.

The invention matching box, there is an inverse conversion FORMED in usually to generate QJ, a moulded 16 to go to the quantization and inverse conversion computer, the invention stipulated with an inverse conversion substrate is used for: a

<img id="idf0007" file="CN100463522CD00161.tif" img-content="drawing" img-format="tif" />

The warning, Q'诸 the groove which is orthogonal, and methods norm is different. With the through groove is composed of a process selection the quantization parameter mm, and upper. Moreover, pays warning FORMED by. Generation, left side, the left. Is a transposition, 1 and 3 bar is multiplied is 1/2 of the flat. The QP value for directly, wherein the D0, D1 and D2 is existed associated of the edge and performs adjustment beforehand, therefore no need to match computing complexity to compensate. Intervals; and matrix of the phase-discrimination is multiplied is 1/2 in only by calculating the inverse flowing from the position, the fact completes through the efficient the cross operation. The be noises for warning; the precise not adjacent that is substantially not affected to the speed for performance.

The LEFT; and stipulated for maximum wherein an equal to 4. Therefore, 2 maintaining-operating work station inverse conversion process's multiplier without expansion 4, therefore of equal to 16 operations.

Digital 2 the inverse cover fixing processing 236 and good inverse conversion processing 238 )for operation output pixel data 240 of the inverse conversion processing result). Hence, by the inverse conversion matrix and transfer (A) BCD special for inverse conversion circuit or, corresponding inverse conversion conductor (a' b' c' d' certain) is a differential: athe u TYPE interface, A.C.

the v; A-c

the y (B &gt;&gt; 1) is d;

the z (D &gt;&gt; 1) as; THE

a/ The u as z; the b of the v y'-c interface, the v of y.

the d'; u-z

, U, v, and y is z the subsidiary variables, wherein the is 1 to refer to the cross 1 (equivalent is multiplied is 1/2). Similar to the code, wherein the disc-shaped structure of shown the digital 4, the formulas the fact reduced to each coefficient to 4 add operations and shape of functions, time 16 operations. Finally, the inverse conversion processing 234 digital (2) is convert restructuring of 2-7 factors the pixel value (completes 1 to 7 the cross operations).

A locating 7, wherein the pixel value of restructuring is 0.01-0.99 9, therefore the inverse conversion of equal to 16 operations. A main (quantization table DO, D1 and D2 is, wherein the adapting to the quantization type end of the computer is 16 operations in over-current. Merits part of the following formed with summarized for the invention conversion and inverse conversion stipulated the operation of compares is provided with H.26L the computer:

<img id="idf0008" file="CN100463522CD00162.tif" img-content="drawing" img-format="tif" />

<img id="idf0009" file="CN100463522CD00171.tif" img-content="drawing" img-format="tif" />

The invention 4, additions l-shaped limiting, and is 1 times 16 in operations the phase-discrimination, and is 1 times doubling end is 4 times 16 in operations in operations 32 the top surface, 1), and is conducted in operations 16 now is very good to understand used for storing the invention with is very remarkable, a special 16 processors. For example, wherein fact, a typical 16 processors A by applying the invention, and a speed which are 1 times improved and decoding.

Moreover, a converting/inverse conversion and compression on the effect of quality of with the test. Coding gain's result (two generally and signal-to-noise ratio reinforced) of: The invention the mapping method and H26L conversion module compares, the data detection result, a consumption without and O.OldB, which is a neglect, wherein fact with possibility to the consumption of leads video signal information is clamped with smaller, to pixel prediction error. Moreover, a conversion of H.26L stipulation, comprising of the total performance and each parameter of 32 quantization parameters, the plurality of ±0.5% is 0.01-0.99 differences, and averaging value the zero, therefore indicated of conversion and H of this invention. the cover 26L stipulated is equally good.

From the detailed description, the claims a a conversion and inverse conversion matrix improved is provided with an image or a video coding and decoding are, ranges from the existing transformations, capable of reducing computing complexity, and a compressing zero-halogen type adverse impact with a quality. Doubling is no need to, it can still obtain the correcting result, wherein an inverse and conversion processing the locating function in the conversion, a quantization and adapting to the quantization parameter value end of the compensation to the phase-discrimination factors. Due to the parameter value; and when code converting operating and decodes adapting to the quantization and conversion operations and 16 in operations alarm (pixel data unit is 9 or not) for.

The invention can with each revisions and structure, as the emitter of the position, and is made of the specification of. The parts; the understand, connected with limiting the invention the wire groove of claims; the top, the invention flowing comprises a and range of the revisions the invention spirit, a polymeric structure and equivalent.