EP3225031A2

EP3225031A2 - Partitionining method and method for partitioning signalling of a coding tree unit

Info

Publication number: EP3225031A2
Application number: EP15822963.3A
Authority: EP
Inventors: Félix Henry
Original assignee: Orange SA
Current assignee: Orange SA
Priority date: 2014-11-27
Filing date: 2015-11-24
Publication date: 2017-10-04
Also published as: CN113207005A; CN107005719B; CN113207005B; US20170324984A1; WO2016083729A3; US11825129B2; EP4311240A3; US10924769B2; EP4311240A2; US20210127141A1; CN107005719A; FR3029333A1; WO2016083729A2

Abstract

The invention relates to a method for encoding at least one image (ICj), including a step of subdividing the image into a plurality of blocks (CTU1, CTU2,..., CTUu,..., CTUS), characterised in that it includes the following steps: subdividing (C2) at least one current block into a first portion and a second portion, the first portion having a rectangular or square shape and the second portion complementing the first portion in the current block, said second portion having a geometric shape with m sides, wherein m > 4; and encoding (C6) the first and second portions.

Description

METHOD FOR ENCODING AND DECODING IMAGES, DEVICE FOR ENCODING AND DECODING AND COMPUTER PROGRAMS

CORRESPONDENTS Field of the invention

The present invention relates generally to the field of image processing, and more specifically to the encoding and decoding of digital images and digital image sequences.

The invention can be applied in particular, but not exclusively, to the video coding implemented in the current AVC and HEVC video coders and their extensions (MVC, 3D-AVC, MV-HEVC, 3D-HEVC, etc.), as well as to the corresponding decoding.

Background of the invention

Current video encoders (MPEG, H.264, HEVC, ...) use a block representation of the images to be encoded. The images are subdivided into blocks of square or rectangular shape, which can be subdivided in turn recursively. In the HEVC standard, such a recursive subdivision follows a tree structure called "quadtree". For this purpose, as shown in Figure 1, a current image IN is subdivided a first time into a plurality of square or rectangular blocks called CTU (Coding Tree Unit), designated CTU _1; CTU ₂ , CTU ,, ..., CTU _L. Such blocks are, for example, 64 × 64 pixels (1 <i L L).

For a given CTU block, it is considered that this block constitutes the root of a coding tree in which:

a first level of sheets under the root corresponds to a first subdivision depth level of the CTU block, for which the CTU block has been subdivided a first time into a plurality of square or rectangular coding blocks called CU (of the English "Coding Unit"),

a second level of sheets under the first level of sheets corresponds to a second level of block partitioning depth CTU, for which some blocks of said plurality of block coding blocks partitioned a first time are partitioned into a plurality of CU type coding blocks, ...

a kth level of sheets under the k-1th level of sheets corresponding to a kth level of partitioning depth of the block CTUi for which some blocks of said plurality of coding blocks of the partitioned block k-1 times are partitioned one last time into a plurality of CU type coding blocks.

In a HEVC-compatible encoder, the partitioning iteration of the CTUi block is performed up to a predetermined partition depth level.

At the end of the aforementioned successive partitioning of the CTU block, as represented in FIG. 1, the latter is finally partitioned into a plurality of coding blocks denoted UC-1, UC ₂ ,. , UC _M with 1 <j≤M.

The purpose of such a subdivision is to delimit zones that adapt well to the local characteristics of the image, such as for example a homogeneous texture, a constant movement, an object in the foreground in the image, etc.

For a CTU block, considered, several different subdivisions of the latter are put in competition with the coder, that is to say respective different combinations of subdivision iterations, in order to select the best subdivision, that is to say ie that which optimizes the coding of the CTU block, considered according to a predetermined coding performance criterion, for example the cost rate / distortion or compromise efficiency / complexity, which are criteria well known to those skilled in the art.

Once the optimal subdivision of a CTU block has been performed, a sequence of digital information, such as, for example, a series of bits, representative of this optimal subdivision, is transmitted in a data signal intended to be stored at the encoder or well transmitted to a video decoder to be read and then decoded by the latter.

In the example of FIG. 1, the binary sequence representative of the optimal subdivision of the CTU block contains the following seventeen bits: 1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0 , 0, 0, 0, 0, 0, 0, for which:

the first bit "1" indicates a subdivision of the block CTU, in four smaller sub-blocks UC-i, UC ₂ , UC ₃ , UC ₄ , the second bit "1" indicates a subdivision of the sub-block UCi into four smaller sub-blocks UC ₅ , CPU ₆ , CPU ₇ , CPU ₈ ,

the third bit "0" indicates a lack of subdivision of the UC ₂ sub-block,

the fourth bit "0" indicates a lack of subdivision of the UC ₃ sub-block,

the fifth bit "0" indicates a lack of subdivision of the sub-block UC ₄ ,

the sixth bit "0" indicates a lack of subdivision of the UCs sub-block,

the seventh bit "1" indicates a subdivision of the sub-block UC ₆ into four smaller sub-blocks UC ₉ , UC- | ₀ , UC-n, UC- | ₂ ,

the eighth bit "1" indicates a subdivision of the sub-block UC ₇ into four smaller sub-blocks UC13, UCi ₄ , UC15, UCi ₆ ,

the ninth bit "0" indicates a lack of subdivision of the UC ₈ sub-block,

the tenth bit "0" indicates a lack of subdivision of the sub-block UC ₉ ,

the eleventh bit "0" indicates a lack of subdivision of the subunit UC ₁₀ ,

- the twelfth bit "0" indicates a lack of subdivision of the

- the thirteenth bit "0" indicates a lack of subdivision of the

- the fourteenth bit "0" indicates a lack of subdivision of the

- the fifteenth bit "0" indicates a lack of subdivision of the

- the sixteenth bit "0" indicates a lack of subdivision of the

the seventeenth bit "0" indicates no subdivision of the UC-I6 sub-block. The bit sequence obtained requires that a sub-block traversal order be predetermined in order to know to which sub-block corresponds a syntax element indicating the subdivision performed. As represented by the arrow F in FIG. 1, such a travel order is generally lexicographic, that is to say that for each level of subdivision considered:

the sub-blocks are traversed starting with the first sub-block UCi located at the top left of the block CTU, and so on until reaching the sub-block UC ₄ located at the bottom right of the block CTU ,,

the sub-blocks resulting from the subdivision of the sub-block UC ₆ are traversed starting with the first sub-block UC ₉ located at the top left of the sub-block UC ₆ and so on until reaching the sub-block UCi ₂ located at the bottom right of the UC ₆ sub-block,

the sub-blocks resulting from the subdivision of the sub-block UC ₇ are traversed starting with the first sub-block UC ₁₃ located at the top left of the sub-block UC ₇ and so on until reaching the sub-block UCi ₆ located at the bottom right of the UC ₇ sub-block.

The aforementioned seventeen bits are written one after the other in the binary sequence which is then compressed by a suitable entropic coding.

For at least one sub-block considered among the different sub-blocks obtained, a prediction of pixels of the sub-block considered is implemented with respect to prediction pixels that belong to either the same image (intra prediction) or to one or more previous images of an image sequence (inter prediction) that have already been decoded. Such prior images are conventionally referred to as reference images and are stored in memory at both the encoder and the decoder. During such a prediction, a residual sub-block is calculated by subtraction of the pixels of the sub-block considered, prediction pixels. The coefficients of the calculated residual sub-block are then quantized after a possible mathematical transformation, for example of the discrete cosine transform (DCT) type, then coded by an entropy coder.

The choice between inter or intra prediction mode is done at the level of each of the UC-i, UC ₂ , ..., UC _{j subunits} , ..., UC _M which can themselves be partitioned into sub-prediction blocks ("prediction units" in English) and transform sub-blocks ("Transform Units"). Each of the prediction sub-blocks and the transform sub-blocks are in turn capable of being sub-divided recursively into sub-blocks according to the abovementioned "quadtree" tree structure.

The block CTUi and its sub-blocks Ud, UC ₂ , ..., UCj, ..., UCM, its prediction sub-blocks and its transform sub-blocks, are likely to be associated with information describing their content.

Such information includes the following:

the prediction mode (intra prediction, inter prediction, prediction by default carrying out a prediction for which no information is transmitted to the decoder ("in English" skip "));

the type of prediction (orientation, reference image component, etc.);

- the type of subdivision into sub-blocks;

the type of transform, for example DCT 4x4, DCT 8x8, etc .;

the pixel values, the transformed coefficient values, the amplitudes, the signs of the quantified coefficients of the pixels contained in the block or sub-block considered.

This information is also recorded in the aforementioned data signal.

When encoding a still image or an image of a sequence of images using sub-block subdivision of quadtree type, it is common to find in the image a significant object of medium or small size which is located in a relatively homogeneous image area. Such a configuration is for example represented in FIG. 2A which represents, as a significant element, a star, which is contained in a homogeneous zone such as, for example, a sky of uniform color.

After implementing a subdivision into quadtree blocks and sub-blocks as shown in FIG. 2B, it is possible to isolate the "star" significant element in a UC ₈ sub-block adapted to its size. A disadvantage of such a subdivision is that it requires the transmission of a binary sequence representative of this subdivision which contains a significant number of bits. Such a sequence proves to be expensive to report, which does not make it possible to optimize the reduction of the compression gain of the coded data. This results in compression performance that is not satisfactory.

Object and summary of the invention

One of the aims of the invention is to overcome disadvantages of the state of the art mentioned above.

For this purpose, an object of the present invention relates to a method of encoding at least one image, comprising a step of subdivision of the image into a plurality of blocks.

The coding method according to the invention is remarkable that it comprises the following steps:

subdividing at least one current block into a first and a second part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the current block, the second part having a geometric shape with m sides with m> 4,

- code the first and second parts.

Such an arrangement makes it possible to very simply subdivide a block into only two parts. The representative binary sequence of this subdivision necessarily contains fewer bits than the representative binary sequence of a "quadtree" type subdivision. The binary sequence representative of the subdivision according to the invention is therefore much less expensive to report.

Furthermore, the subdivision according to the invention is particularly well suited to the case where blocks of the image contain a significant element, for example an object in the foreground, which is located in a homogeneous zone having a low energy, such as for example a background of color, orientation or uniform movement. Correlatively, the invention relates to a device for coding at least one image, comprising a partitioning module for subdividing the image into a plurality of blocks.

Such a coding device is remarkable in that the partitioning module is able to subdivide at least one current block into a first and a second part, the first part having a rectangular or square shape and the second part forming the complement of the first part. part in the current block, the second part having a geometric shape with m sides, with m> 4,

and in that it comprises a coding module for coding the first and second parts.

Correspondingly, the invention also relates to a method of decoding a data signal representative of at least one coded picture having been subdivided into a plurality of blocks.

Such a decoding method is remarkable in that it comprises the following steps:

- decode the first and second parts.

Such an arrangement makes it possible to very simply subdivide a current block to be decoded into only two parts, such a subdivision being much less complex to perform than a "quadtree" type subdivision.

Moreover, the subdivision according to the invention is particularly well suited to the case where blocks of the image to be decoded contain a significant element, for example an object in the foreground, which is located in a homogeneous zone having a low energy, such as for example a background of color, orientation or uniform movement.

In a particular embodiment, during the step of decoding the second part with m sides of the current block, at least one information for reconstructing the pixels of the second part with m sides of the current block is set to a predetermined value. . An advantage of such an arrangement lies in the fact that the decoder autonomously determines said at least one pixel reconstruction information of the second part with m sides. In other words, said at least one corresponding reconstruction information is advantageously not transmitted in the data signal received at the decoder. Thus the reduction of the signaling cost is optimized.

According to a variant, said at least one information for reconstructing the pixels of the second part with m sides of the current block is representative of the absence of subdivision of the second part with m sides of the current block.

Advantageously, when decoding the second part m-sides of the current block, the decoder autonomously determines that it does not need to subdivide this part, since it characterizes a homogeneous region of the current block to be decoded. is devoid of details.

According to another variant, said at least one information for reconstructing the pixels of the second part with m sides of the current block is representative of the absence of residual information resulting from a prediction of the pixels of the second part with m sides of the block. current.

Advantageously, when decoding the second part of the current block, the decoder autonomously determines that the residual pixels obtained as a result of the prediction of said second part with m sides have a zero value. It is considered that the second part with m sides is associated with a zero prediction residue since it characterizes a homogeneous region of the current block to be decoded.

According to another variant, said at least one information for reconstructing the pixels of the second part with m sides of the current block is representative of predetermined prediction values of the pixels of the second part with m sides of the current block.

Such a variant makes it possible to further optimize the signaling cost by avoiding transmitting in the data signal the index of the prediction mode which has been selected to the coding in order to predict the second part with m sides of the current block.

In another particular embodiment, the decoding method comprises, prior to the step of subdivision of the current block, a step reading, in the data signal, information indicating whether the current block is intended to be subdivided into a first and a second part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the current block, the second part having a geometric shape with m sides, with m> 4, or to be subdivided according to another predetermined method.

Such an arrangement enables the decoder to determine whether, during coding of a current block, the coder has activated or not the subdivision of the current block according to the invention, for a sequence of images considered, for an image considered or else for a portion of image ("slice" in English) considered, so that the decoder can implement correspondingly subdivision performed coding. As a result, such a decoding method is particularly flexible because adaptable to the current video context. Indeed, the decoding method is adapted to implement the subdivision according to the invention or according to another type of subdivision, such as for example the quadtree subdivision, as a function of the value taken by a dedicated indicator inscribed in the signal of data.

Such a dedicated indicator remains relatively compact to report and allows to maintain the compression gain obtained through the subdivision according to the invention.

In yet another particular embodiment, the decoding method comprises a step of reading, in the data signal, information indicating a subdivision configuration of the current block selected from different predetermined subdivision patterns.

Such an arrangement makes it possible to adapt the subdivision according to the invention as a function of the location of the significant element in the homogeneous region of the current block.

In yet another particular embodiment, the step of decoding the second part of the current block with m sides comprises the substeps consisting of:

to apply an entropy decoding to the pixels of the second part with m sides, - Complement the decoded pixels entropically of the second part to m sides by pixels reconstructed according to a predetermined reconstruction method, until a block of square or rectangular pixels.

Such an arrangement advantageously makes it possible, when a step of application of a transform is to be implemented following the entropy decoding step of the second part of the current block to be decoded, to reuse the hardware tools. and square or rectangular block transform software that is commonly implemented in current video encoders and decoders.

In yet another particular embodiment, a subdivided current block contains at most a portion having a geometric shape with m sides.

Such an arrangement is well suited to the case where the current block contains two distinct zones of texture, that is to say that defined by a single significant element and that defined by a single homogeneous zone. Advantageously, it is therefore not necessary to re-subdivide the current block to be decoded.

The various embodiments or aforementioned embodiments can be added independently or in combination with each other, to the steps of the decoding method as defined above.

Correlatively, the invention relates to a device for decoding a data signal representative of at least one coded picture having been subdivided into a plurality of blocks.

Such a decoding device is remarkable in that it comprises:

a partitioning module for subdividing at least one current block into a first and a second part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the current block, the second part having a geometric shape with m sides, with m> 4,

a decoding module for decoding the first and second parts. The invention also relates to a computer program comprising instructions for implementing one of the coding and decoding methods according to the invention, when it is executed on a computer.

Such a program can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any another desirable form.

Still another object of the invention is directed to a computer readable recording medium, and including computer program instructions as mentioned above.

The recording medium may be any entity or device capable of storing the program. For example, the medium may include storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or a magnetic recording means, for example a USB key or a hard disk.

On the other hand, such a recording medium may be a transmissible medium such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means. The program according to the invention can be downloaded in particular on an Internet type network.

Alternatively, such a recording medium may be an integrated circuit in which the program is incorporated, the circuit being adapted to execute the method in question or to be used in the execution of the latter.

Brief description of the drawings

Other features and advantages will appear on reading several preferred embodiments described with reference to the figures in which:

FIG. 1 represents an example of a conventional subdivision of a current block, such as the "quadtree" subdivision,

FIGS. 2A and 2B show an application of the "quadtree" type subdivision to a current block containing a single element significant, a star, which is contained in a homogeneous zone such as for example a sky of uniform color,

FIG. 3 represents the main steps of the coding method according to one embodiment of the invention,

FIG. 4 represents an embodiment of a coding device according to the invention,

FIGS. 5A to 5J respectively represent various subdivision embodiments according to the invention of the current block,

FIGS. 6A and 6B respectively represent two coding embodiments of the parts obtained by subdivision of the current block, according to a type of subdivision represented in FIG. 5A,

FIG. 7 represents an example of subdivision of the current block to which the coding embodiment of FIG. 6B applies,

FIG. 8 represents the main steps of the decoding method according to one embodiment of the invention,

FIG. 9 represents an embodiment of a decoding device according to the invention,

FIGS. 10A and 10B respectively represent two embodiments of decoding the parts obtained after reconstruction of the subdivision of the current block, according to a type of subdivision shown in FIG. 5A.

Detailed description of the coding part

An embodiment of the invention will now be described, in which the coding method according to the invention is used to code an image or a sequence of images according to a binary signal close to that obtained by a coding set. implemented in an encoder according to any current or future video coding standards.

In this embodiment, the coding method according to the invention is for example implemented in a software or hardware way by modifications of such an encoder. The coding method according to the invention is represented in the form of an algorithm comprising steps C1 to C7 as represented in FIG. According to the embodiment of the invention, the coding method according to the invention is implemented in a coding device or coder CO shown in FIG.

As illustrated in FIG. 4, such an encoder comprises a memory MEM_CO comprising a buffer memory TAMP_CO, a processing unit UT_CO equipped for example with a microprocessor μΡ and driven by a computer program PG_CO which implements the coding method according to the invention. At initialization, the code instructions of the computer program PG_CO are for example loaded into a RAM memory (not shown) before being executed by the processor of the processing unit UT_CO.

The coding method represented in FIG. 3 applies to any current image IC _{j that is} fixed or part of a sequence of L images ld, IC _j ,..., IC _L (1 _{j j} _{L L} ) to be encoded. .

During a step C1 shown in FIG. 3, in a manner known per se, a current image IC _{j is} subdivided into a plurality of blocks of the above-mentioned CTU type: CTU _1; CTU ₂ , CTU _U , ..., CTU _S (1 <u _{S S} ).

Such a subdivision step is implemented by a processor or partitioning software module MP_CO shown in FIG. 4, which module is driven by the microprocessor μΡ of the processing unit UT_CO.

Preferably, each of CTU-i, CTU ₂ , CTU _U , ..., CTUs has a square shape and comprises NxN pixels, with N> 2.

According to an alternative, each of the CTU-i, CTU ₂ , CTU _U , ..., CTU blocks has a rectangular shape and comprises NxP pixels, with N> 1 and P> 2.

During a step C2 shown in FIG. 3, for a current block

CTU _U previously selected, the partitioning module MP_CO of Figure 4 subdivides the current block CTU _U into at least a first part and a second part, the first and second parts being complementary to one another. According to the invention:

the first part has a rectangular or square shape,

- and the second part has a geometric shape with m sides, with m> 4. According to a preferred embodiment, the current block CTU _U is subdivided:

in a first portion of rectangular or square shape or in a plurality of parts of rectangular or square shape,

and in at most a second portion of geometric shape with m sides.

Within the meaning of the invention, the first and second parts respectively form two separate coding units CUi and CU ₂ . This last terminology is notably used in the standard HEVC "ISO / IEC / 23008-2 Recommendation ITU-T H.265 High Efficiency Video Coding (HEVC)".

According to a first embodiment of subdivision shown in FIG. 5A, for a current block CTU _U square of size NxN:

the first part CUi is a square block of size ^ x ^,

and the second part CU ₂ , which forms the complement of the first part CUi in the current block CTU _U , has a geometric shape with m sides, with m> 4.

In the example shown in FIG. 5A, m = 6.

As shown in FIG. 5A, four types of subdivision SUB1 S1, SUBS1 SUB4 ₁ of the current block CTU _U are possible, the square block CU ₁ being able to be located in one of the four corners of the current block CTU _U.

For the sake of clarity in FIG. 5A, only in the case of, for example, subdivision types SUB1 and SUB3i, are represented:

the first pixel ptl _u of the current block CTU _U , with coordinates (xmin, ymin), which is located at the top left of the latter,

the last pixel pbr _u of the current block CTU _U , of coordinates (max, y _m ax), which is located at the bottom and to the right in the latter,

the first pixel pt of the first part CU-i, with coordinates (x'min, y'min), which is located at the top left of the latter,

the last pixel pbr- of the first part CU-i, of coordinates (x'max, y'max) which is located at the bottom and to the right in the latter. According to the particular type of subdivision SUBI -1, the first pixel ptl _u of the current block CTU _U is the same as the first pixel pth of the first part Clh.

Whatever the type of subdivision chosen, the second part ₂ of geometric form m-sides is then generally defined as a set of pixels ptl ₂ such that for any pixel pv ₂ (x " _v , y" v) of this set :

Xmin- Xv- Xmax St ymin- y v-ymax

. X " _v <X'min OR X" _v >X'max OR yV min OR y " _v >y'max

According to a second embodiment of subdivision shown in FIG. 5B, for a current block CTU _U square of size NxN:

the first part CU ₁ is a square block of size ^ x ^,

and the second part CU ₂ , which forms the complement of the first part CU ₁ in the current block CTU _U , has a geometric shape with m sides, with m> 4.

In the example shown in FIG. 5B, m = 6 or m = 8.

As shown in FIG. 5B, sixteen subdivision types SUB1 ₂ , SUB2 ₂ , SUB16 ₂ of the current block CTU _U are possible, the square block CU ₁ being able to be located in sixteen different positions inside the current block CTU _U , by successive translation of N / 4 pixels of the square block CU ₁ inside the current block CTU _U.

For the sake of clarity in FIG. 5B, only in the case, for example, of subdivision types SUB1 ₂ and SUB9 ₂ , are represented:

the first pixel pt1 _u of the current block CTU _U , with coordinates (Xmin, ymin), which is located at the top left of the latter,

the last pixel pbr _u of the current block CTU _U , of coordinates (xmax, ymax), which is situated at the bottom and to the right in the latter,

the first pixel pth of the first part Clh, with coordinates (x'min, y'min), which is situated at the top and left in the latter,

the last pixel pbn of the first part Clh, of coordinates (x'max, y'max) which is located at the bottom and to the right in the latter. According to the particular subdivision mode SUB1 ₂ , the first pixel pt1 _u of the current block CTU _U is the same as the first pixel pt of the first part

Whatever the type of subdivision chosen, the second part ₂ of geometric shape m-side is then defined generally as a set of pixels such that for any pixel pv ₂ (x " _v , y" v) of this set :

Xmin- Xv- Xmax St ymin- y v-ymax

. X " _v <X'min OR X" _v >X'max OR yV min OR y " _v >y'max

According to third, fourth, fifth and sixth subdivision embodiments respectively represented in FIGS. 5C, 5D, 5E and 5F, for a CTU _U square current block of size NxN:

- the first CUi part is a rectangular block size UXV pixels, such that U <N and V <N, all of the coordinates of such a rectangular block being selected from a predefined _list LT several sets of coordinates defining each a rectangular block of a predetermined shape, the list LT _a being stored in the buffer memory TAMP_CO of the encoder CO of FIG. 4,

In each of FIGS. 5C to 5F, only one type of subdivision of the current block CTU _U has been represented, although of course there may be several of them.

In addition, the definition of the second portion CU ₂ of the current block CTU _U is the same as that given in the examples of FIGS. 5A and 5B.

According to seventh, eighth, ninth and tenth subdivision embodiments respectively shown in Figs. 5H, 51, 5J and 5K, the current block CTU _U is a rectangle of size NxP pixels, with N> 1 and P> 2.

According to these four modes of subdivision:

the first part CUi is a rectangular block of size UxV, such that U <N and V <P, the set of coordinates of such a rectangular block being chosen from a list LT _b predefined of several sets of coordinates each defining a rectangular block of a predetermined shape, the list LT _b being stored in the buffer TAMP_CO of the coder CO of FIG. 4,

In each of FIGS. 5G to 51, only one type of subdivision of the current block CTU _U has been shown, although of course there may be several.

During a step C3 shown in FIG. 3, each of the CTU current blocks _U , or only a portion, which has been subdivided in accordance with the different subdivision modes according to the invention as represented in FIGS. 5A to 5K, is set in competition :

with different current CTU _U blocks respectively subdivided according to different well-known subdivision modes, such as for example subdivided into only four rectangular or square blocks, subdivided according to the "quadtree" method, etc.

and with a CTU _U non-subdivided current block.

Such placing in competition is performed according to a predetermined coding performance criterion of the CTU _U current block, for example the cost rate / distortion or a compromise between efficiency and complexity, which are criteria well known to those skilled in the art.

The competition is implemented by a processor or calculation software module CAL1_CO shown in FIG. 4, which module is driven by the microprocessor μΡ of the processing unit UT_CO.

At the end of the competition, an optimum subdivision mode SUBo _pt of the current block CTU _U is selected, that is to say it is the one that optimizes the coding of the block CTU _U by minimization cost / distortion cost or by maximizing the efficiency / complexity trade-off.

During a step C4 represented in FIG. 3, an indicator representative of the subdivision mode selected at the end of step C3 is selected in a correspondence table TC stored in the buffer TAMP_CO of the coder CO of FIG. 4.

Such a selection is implemented by a processor or calculation software module CAL2_CO shown in FIG. 4, which module is driven by the microprocessor μΡ of the processing unit UT_CO.

The indicator representative of a given subdivision mode is for example a syntax element called cut_type which, according to a preferred embodiment, takes for example three values:

- 0 to indicate a conventional subdivision of the current block into four rectangular or square blocks,

- 1 to indicate a subdivision of the current block in accordance with the subdivision mode shown in FIG. 5A,

- 2 to indicate a subdivision of the current block according to the subdivision mode shown in FIG. 5B,

- 3 to indicate a lack of subdivision of the current block.

On the other hand, in the case where the syntax element type_decut is 1, the latter is associated, in the correspondence table TC of FIG. 4, with another syntax element called arr_decoupe1 which indicates the type of subdivision SUBI -i, SUB2i, SUB3i, SUB4i selected, as shown in Figure 5A. The syntax element arr_decoupe1 takes the value:

- 0 to indicate the subdivision type SUB1

- 1 to indicate the type of subdivision SUB2i,

- 2 to indicate the type of subdivision SUB3i,

- 3 to indicate the type of subdivision SUB4.

On the other hand, in the case where the syntax element type_decut is 2, the latter is associated, in the correspondence table TC of FIG. 4, with another syntax element called arr_decoupe2 which indicates the type of subdivision chosen from the sixteen subdivision types SUB1 ₂ , SUB2 ₂ , SUBI 6 ₂ of the current block CTU _U , as shown in FIG. 5B. The syntax element arr_decoupe2 has the value:

- 0 to indicate the subdivision type SUB1 ₂ ,

- 1 to indicate the subdivision type SUB2 ₂ ,

- 2 to indicate the type of subdivision SUB3 ₂ , - 3 to indicate the type of subdivision SUB4 ₂ ,

- 4 to indicate the subdivision type SUB5 ₂ ,

- 5 to indicate the type of subdivision SUB6 ₂ ,

- 6 to indicate the subdivision type SUB7 ₂ ,

- 7 to indicate the subdivision type SUB8 ₂ ,

- 8 to indicate the subdivision type SUB9 ₂ ,

- 9 to indicate the type of subdivision SUB10 ₂ ,

- 10 to indicate the subdivision type SUB1 1 ₂ ,

- 1 1 to indicate the subdivision type SUB12 ₂ ,

- 12 to indicate the type of subdivision SUB13 ₂ ,

- 13 to indicate the type of subdivision SUB14 ₂ ,

- 14 to indicate the subdivision type SUB15 ₂ ,

- 15 to indicate the type of subdivision SUB16 ₂ .

During a step C5 shown in FIG. 3, the value of the syntax element type_cut which has been selected at the end of the above-mentioned step C4 is coded and, if applicable , to the coding of the syntax element arr_decoupe1 or arr_decoupe2 associated with it.

The above-mentioned step C5 is implemented by a processor or indicator coding software module MCI as represented in FIG. 4, which module is controlled by the microprocessor μΡ of the processing unit UT_CO.

During a step C6 shown in FIG. 3, the portions CUi and CU ₂ of the current block CTU _{U are} coded according to a predetermined order of travel. According to a preferred embodiment, the first part CUi is coded before the second part CU ₂ . As an alternative, the first part CUi is coded after the second part CU ₂ .

The coding step C6 is implemented by a processor or UCO coding software module as shown in FIG. 4, which module is driven by the microprocessor μΡ of the processing unit UT_CO.

As shown in more detail in FIG. 4, the UCO coding module conventionally comprises:

a processor or prediction software module PRED_CO,

a processor or software module for calculating residual data CAL3_CO, a processor or software module MT_CO of transformation of the DCT type (abbreviation of "Discrete Cosine Transform"), DST (abbreviation of "Discrete Sine Transform"), DWT (abbreviation of "Discrete Wavelet Transform")

a processor or quantification software module MQ_CO,

a processor or software module MCE_CO of entropy coding, for example of CABAC type ("Context Adaptive Binary Arithmetic Coder" in English) or a Huffman coder known as such.

During a step C7 shown in Figure 3, it is proceeded to the construction of a data signal F which contains the data coded at the end of steps C5 and C6 above. The data signal F is then transmitted by a communication network (not shown) to a remote terminal. This includes a decoder which will be described later in the following description.

Step C7 is implemented by a processor or software module MCF for constructing a data signal, as shown in FIG. 4.

The coding steps which have just been described above are implemented for all the CTU-i, CTU ₂ , CTU _U , ..., CTUs blocks to be encoded of the current image IC _j considered, in one order predetermined which is for example the lexicographic order.

Other types of course than the one just described above are of course possible.

FIG. 6A will now describe a first embodiment of the different substeps implemented during the aforementioned coding step C6 in the UCO coding module shown in FIG. 4.

According to this first embodiment, the optimal subdivision mode SUB _op t which has been selected at the end of coding step C3 is for example one of the subdivision modes shown in FIG. 5A. For this purpose, it is the indicator type_cut of value 1 which was selected at the end of step C4 above. More specifically, it is for example the type of subdivision SUBD2, as shown in FIG. 5A, which has been selected at the end of the coding step C3. For this purpose, the indicator type_cut of value 1 is further associated with the indicator arr_cutter1 of value 1, as defined above in the description.

The value 1 of the cut_type_indicator is written in compressed form in the data signal F, followed by the value 1 of the cut_stop1 flag.

Moreover, in accordance with the first embodiment of FIG. 6A, the portions CU ₁ and CU ₂ of the current block CTU _U are not subdivided again.

For this purpose, according to one embodiment:

in association with the coded data of the first part CU-i, is associated the indicator type_cut of value 3,

in association with the coded data of the second part CU ₂ , is associated the indicator type_decoupe of value 3.

According to the invention, the value of the cut_type_indicator associated with the coded data of the second portion CU ₂ is written in compressed form in the data signal F before the value of the cutoff_type_indicator associated with the coded data of the first portion CU -i.

In the example of FIG. 6A, the data signal F therefore contains the following values: 1 133 which are representative of the partitioning of the current block CTU _U.

As a variant, considering the fact that the second part CU ₂ defines a homogeneous area of the current block CTU _U , no indicator representative of the absence of subdivision of the part CU ₂ is inscribed in the data signal F represented in FIGS. 3 and 4. According to such a variant, it is indeed assumed in coding, as in decoding, that a part with m sides of the current block is systematically not subdivided. Thus, the transmission to the decoder of a type_decoupe indicator of value 3 is not necessary.

The data signal F therefore contains the following values: 1 13, which reduces the cost of signaling. During a sub-step C610 represented in FIG. 6A, the coding module UCO selects, as current portion CUk (k = 1 or k = 2), either first the square portion CU-i, or of firstly the part CU ₂ with m sides.

During a sub-step C61 1 represented in FIG. 6A, the module PRED_CO of FIG. 4 proceeds to the predictive coding of the current part.

Conventionally, the pixels of the CUi portion are predicted with respect to pixels that have already been coded and then decoded, by putting in competition known intra and / or inter prediction techniques.

Among the possible predictions for the current part CU-i, the optimal prediction is chosen according to a rate-distortion criterion well known to those skilled in the art.

Said aforementioned predictive coding sub-step makes it possible to construct a predicted part CUpi which is an approximation of the current part CU-i. The information relating to this predictive coding will subsequently be entered in the data signal F represented in FIGS. 3 and 4. Such information notably comprises the type of prediction (inter or intra), and, if appropriate, the intra or intra prediction mode. well the reference image index and the displacement vector used in the inter prediction mode. This information is compressed by the coder CO shown in FIG.

During a substep C612, the calculation module CAL3_CO of Figure 4 proceeds to subtract the predicted portion CUpi of the current portion CUi to produce a residue portion CUn.

During a sub-step C613 represented in FIG. 6A, the MT_CO module of FIG. 4 proceeds with the transformation of the residue part CUn according to a conventional direct transformation operation, such as, for example, a discrete cosine transformation of the type DCT, to produce a transformed part CUt-i.

During a sub-step C614 represented in FIG. 6A, the module MQ_CO of FIG. 4 proceeds with the quantization of the transformed part CUti according to a conventional quantization operation, such as, for example, scalar quantization. A portion CUq-ι, formed of quantized coefficients, is then obtained. During a substep C615 shown in FIG. 6A, the MCE_CO module of FIG. 4 proceeds with the entropic coding of the quantized coefficients CUq-i.

The aforementioned sub-steps C61 1 to C615 are then iterated in order to code the second part CU ₂ with m sides of the current block CTU _U.

In accordance with the invention, in the case of encoding the second m-side portion CU ₂ , one or more coding information of the pixels of the second portion CU ₂ are set to predetermined values.

Thus, according to a preferred embodiment variant, during the sub-step C61 1 of predictive coding of the portion CU ₂ of the current block CTU _U , the pixels of the portion CU ₂ are predicted with respect to pixels of corresponding values respectively. predetermined. Such values are stored in a list LP contained in the buffer memory TAMP_CO of the coder CO of FIG. 4.

Preferably, these predetermined prediction values are selected so that during the sub-step C612 of FIG. 6A, the subtraction of the predicted portion CUp ₂ of the current portion CU ₂ produces a residue portion CUr ₂ which includes pixel values that are null or close to zero.

Such an arrangement makes it possible advantageously to take advantage of the homogeneity of the portion CU ₂ of the current block CTU _U while at the same time making it possible to substantially reduce the cost of signaling the coding information of the current block CTU _U in the data signal F.

As a variant, the pixels of the portion CU ₂ are conventionally predicted, in the same way as the portion CU-i.

According to another preferred embodiment, the quantized coefficients of the quantized residual portion CUq ₂ obtained at the end of the substep C614 of FIG. 6A are all set to zero and are not written in the data signal F.

Such an arrangement makes it possible advantageously to take advantage of the homogeneity of the portion CU ₂ of the current block CTU _U while at the same time making it possible to substantially reduce the cost of signaling the coding information of the current block CTU _U in the data signal F. According to the invention, between the substeps C612 and C613 mentioned above, an intermediate substep C6120 is implemented. During this intermediate sub-step, the residual pixels of the residue portion CUr ₂ with m sides are completed by pixels of predetermined respective value, until a block of square or rectangular pixels is obtained.

According to various possible embodiments, the residual pixels of the residue portion CUr ₂ can be completed:

by pixels of respective zero value,

by pixels reconstructed conventionally by interpolation, by pixels reconstructed conventionally using the so-called "inpaiting" technique.

The aforementioned sub-step C6120 is implemented by a processor or calculation software module CAL4_CO as shown in FIG. 4, which module is driven by the microprocessor μΡ of the processing unit UT_CO.

Such an arrangement makes it possible to reuse the transformation module

MT_CO of Figure 4 which conventionally applies square or rectangular block transforms.

Taking into account the fact that sub-step C612 only applies to the second part CU ₂ of geometrical shape with m sides, this sub-step, as well as the calculation module CAL4_CO, are represented in dotted lines, respectively on the Figures 3 and 4.

A second embodiment of the different substeps implemented during the aforementioned coding step C6 in the UCO coding module shown in FIG. 4 will now be described with reference to FIG. 6B.

This second embodiment differs from that of FIG. 6A in that the first portion CUi of the CTU current block _U is subdivided again. An example of such a subdivision of the current block CTU _U is shown in FIG. 7.

In the example of FIG. 7, the optimum subdivision mode SUB _op t which has been selected at the end of the coding step C3 mentioned above is, for example, again the indicator type_cut of value 1 which has been selected at the outcome of step C4 above. As shown in Figure 7, this value is entered in compressed form in the data signal F. As explained above, the indicator type_cut of value 1 is further associated with the indicator arr_cutter1 of value 1, as defined above in the description. As shown in FIG. 7, the value of the indicator arr_cutter1 of value 1 is then written in compressed form in the data signal F following the value of the indicator type_cut.

According to the second embodiment of FIG. 6B, in the same way as in the embodiment of FIG. 6A, the second portion CU ₂ of the current block CTU _U is not subdivided again on the assumption that it is representative of a homogeneous zone of the CTU _U current block.

In association with the coded data of the second portion CU ₂ , the value 3 of the indicator type_cut is written in compressed form in the data signal F, following the value 1 of the indicator arr_cut1. This value is shown in bold in Figure 7.

According to the invention, the value of the type_decoupe flag associated with the coded data of the second part CU ₂ is written in compressed form in the data signal F systematically before the value of the indicator type_decoupe associated with the coded data of the first part CU-i.

As a variant, the value of the cut_type_indicator associated with the coded data of the second portion CU ₂ could be written in compressed form in the data signal F systematically after the value of the indicator type_decoupe associated with the coded data of the first part.

In the example of FIG. 7, the portion CUi is subdivided for example into four square blocks CU1 -i, CU2, CU3i, CU4, according to a conventional subdivision method, of the "quadtree" type, for example.

The coded data of the part CU ₁ are therefore associated with the indicator type_cut of value 0, representative of such a subdivision, as defined above in the description. As shown in FIG. 7, this value is written in compressed form in the data signal F, following the value 3 of the type_decoupe flag.

In the example of FIG. 7, the block CU1 ₁ is not subdivided. The coded data of the CUi part are therefore associated in addition with the indicator type_cut of value 3, representative of the absence of such a subdivision, as defined above in the description. As shown in FIG. 7, this value is written in compressed form in the data signal F, following the value 0 of the type_decoupe flag.

In the example of FIG. 7, the block CU2 is subdivided according to the invention, in particular according to the type of subdivision SUB6 ₂ represented in FIG. 5B. Thus, the block CU2 is subdivided into a first part CU21 of square shape and a second part CU22i with m sides. In the example shown, the second portion CU22i has 8 sides.

The coded data of the CUi part are therefore associated in addition to the indicator type_cut of value 2, itself associated with the indicator arr_decoupe2 of value 6, as defined above in the description. As shown in FIG. 7, these values 2 and 6 are written successively in compressed form in the data signal F, following the value 3 of the type_decoupe flag.

In the example of FIG. 7, the block CU3i is subdivided into four square blocks CU31 i, CU321, CU331, CU341, according to a conventional subdivision method, of the "quadtree" type, for example.

The coded data of the CUi part are therefore associated in addition with the indicator type_cut of value 0, representative of such a subdivision, as defined above in the description. As shown in FIG. 7, this value is written in compressed form in the data signal F, following the value 6 of the indicator arr_decoupe2.

In the example of FIG. 7, the block CU4 is not subdivided.

The coded data of the CUi part are therefore associated in addition with the indicator type_cut of value 3, representative of the absence of such a subdivision, as defined above in the description. As shown in FIG. 7, this value is written in compressed form in the data signal F, following the value 0 of the type_decoupe flag.

The second part CU22i of the block CU2i is not subdivided again on the assumption that it is representative of a homogeneous zone of this block. In association with the coded data of the first portion CU-i, the value 3 of the indicator type_cut is then written in compressed form in the data signal F, following the value 3 of the indicator type_decoupe. This value is shown in bold in Figure 7.

According to the invention, the value of the cut_type_indicator associated with the CU22 part m side of the CU2 block is written in compressed form in the data signal F systematically before the value of the cut_type_indicator associated with the square portion CU21 of the CU2-I block.

As a variant, the value of the indicator type_decoupe associated with the part

CU22-I with m sides of the block CU2 could be written in compressed form in the data signal F systematically after the value of the indicator type_decoupe associated with the square portion CU21 ₁ of the block CU2-I.

In the example of FIG. 7, the first portion CU21 i of the block CU2i is not subdivided. In association with the coded data of the first part CU-i, the value 3 of the indicator type_decoupe is then written in compressed form in the data signal F, following the value 3 of the indicator type_decoupe associated with the CU22i part with m sides of the CU2-I block.

In the example of FIG. 7, the four CU31 blocks i, ΰΙΙ32 _1; CU33i, CU34-I of the CU3i block are not subdivided. The value 3 of the cut_type_indicator is then written in compressed form successively four times in the data signal F, following the value 3 of the cut_type_indicator associated with the CU21 part ₁ of the CU2-I block.

As a variant to this second embodiment, the two values 3 of the indicator type_decoupe as represented in bold in FIG. 7 and representative of the absence of subdivision of the m-side portions CU ₂ and CU22i of the current block CTU _U do not are not written in the data signal F, which reduces the cost of signaling. It is indeed assumed in coding, as in decoding, that a part with m sides of the current block is systematically not subdivided. Thus, the transmission to the decoder of a type_decoupe indicator of value 3 is not necessary.

Reference is again made to Figure 6B. During a sub-step C620 represented in FIG. 6B, the coding module UCO selects, as current portion CU _k (k = 1 or k = 2), either first of all the square portion CU-i, or of firstly the part CU ₂ with m sides.

During a sub-step C621 shown in FIG. 6B, the UCO coding module tests whether the index k associated with the current portion CU _k is 1 or 2.

If the index k is equal to 2, the portion CU ₂ of the current block CTU _U is coded according to the substeps C61 1 to C615 of FIG. 6A.

If the index k is equal to 1, during a sub-step C622 shown in FIG. 6B, the UCO coding module of FIG. 4 selects a current sub-portion CU _k 'of the first portion CUi of the block current CTU _U , such that 1 <k'≤N.

In the example shown in FIG. 7, N = 8, since the first portion CUi of the current block CTU _U has been subdivided into eight sub-parts of the "coding unit" type CU1 i, CU21 _1; CU22 _1; CU31 i, CU32 _1; CU33i, CU34 _1; CU4i.

During a substep C623 shown in FIG. 6B, the module

PRED_CO of FIG. 4 selects for this current subpart CU an intra or intra prediction mode, for example by putting these modes in competition according to a rate-distortion criterion. The prediction mode selected is associated with an indicator _PR which is intended to be transmitted in the data signal F.

During an optional substep C624 shown in FIG. 6B, the partitioning module MP_CO of FIG. 4 subdivides the current subpart CU _k into a plurality W of prediction sub-parts PU-i, PU ₂ , PU _Z , ... PU _W (1≤z≤W) of the type "prediction unit" above. Such a subdivision may be conventional or in accordance with the invention, as shown in FIGS. 5A and 5B. In a manner similar to what has been described with reference to the embodiment of FIG. 6A, a succession of indicators representative of the subdivision is intended to be transmitted in the data signal F.

During an optional sub-step C625 shown in FIG. 6B, the UCO coding module of FIG. 4 selects a first current sub-portion PU _Z. Such a selection is made in a predefined order, such as for example the lexicographic order. During an optional sub-step C626 represented in FIG. 6B, the module PRED_CO of FIG. 4 selects for the current subpart PU _Z the optimal prediction parameters associated with the prediction mode selected in the substep C623 mentioned above. . If, for example, the prediction mode INTER has been selected in the substep C623 mentioned above, the optimal prediction parameters are one or more motion vectors, as well as one or more reference images, such optimal parameters making it possible to obtain the best coding performance of the current subpart PU _Z according to a predetermined criterion, such as for example the rate-distortion criterion. If, for example, the INTRA prediction mode has been selected in the aforementioned sub-step C623, the optimal prediction parameters are associated with an INTRA mode selected from among various available INTRA modes. As for the INTER mode, the optimal prediction parameters are those that make it possible to obtain the best coding performance of the current sub-part PU _Z according to a predetermined criterion, such as, for example, the rate-distortion criterion.

The substeps C625 to C626 are iterated for each of the sub-parts PU-1, PU ₂ , PUz,... PUw of the current subpart CUk 'of the first part CUi of the current block CTU _U , in the predetermined lexicographic order.

During an optional sub-step C627 shown in FIG. 6B, the partitioning module MP_CO of FIG. 4 sub-divides the current subpart CU CU into a plurality Z of transform subparts TU-i, TU ₂ , TU _W , ... TU _Z (1≤w≤Z) of the aforementioned "transform unit" type. Such a subdivision may be conventional or in accordance with the invention, as shown in FIGS. 5A and 5B. In a manner similar to what has been described with reference to the embodiment of FIG. 6A, a succession of indicators representative of the subdivision is intended to be transmitted in the data signal F.

During an optional sub-step C628 shown in FIG. 6B, the UCO coding module of FIG. 4 selects a first sub-part of the current transform TU _W. Such a selection is made in a predefined order, such as for example the lexicographic order. During a sub-step C629 represented in FIG. 6B, the calculation module CAL3_CO of FIG. 4 proceeds, in a manner similar to the sub-step C612 of FIG. 6A, to the calculation of a sub-part residue TUr _w .

During a sub-step C630 represented in FIG. 6B, the MT_CO module of FIG. 4 transforms the residue sub-part TUr _w according to a conventional direct transformation operation, such as, for example, a cosine transformation. discrete DCT type, to produce a transformed sub-part TUt _w .

During a sub-step C631 represented in FIG. 6B, the module MQ_CO of FIG. 4 proceeds with the quantization of the transformed sub-part TUtw according to a conventional quantization operation, such as for example a scalar quantization. A subset TUq _w , formed of quantized coefficients, is then obtained.

During a sub-step C632 represented in FIG. 6B, the MCE_CO module of FIG. 4 proceeds to the entropy coding of the quantized coefficients TUq _w .

The set of substeps C628 to C632 is iterated for each of the subsections TU-i, TU ₂ , TU _W , ..., TUz of the current subpart CUk 'of the first part CUi of the current block CTU _U , in the predetermined lexicographic order.

According to the invention, in the case where the current transformed sub-part TU _W has a geometric shape with m sides, an intermediate substep C6290 is implemented between the sub-steps C629 and C630 mentioned above. During this intermediate sub-step, the residual pixels of the residual sub-part TUr _w to m sides are completed by pixels of zero value or coded according to a predetermined coding method, until a block is obtained. square or rectangular pixels.

The aforementioned sub-step C6290 is implemented by the calculation software module CAL4_CO as represented in FIG. 4.

If the sub-step of calculation C6290 is implemented, during the sub-step C631 represented in FIG. 6B, the module MQ_CO of FIG. 4 proceeds to the quantification of the transformed current subpart TUt _w to the exclusion of the pixels added during the substep C6290 and which were transformed during the substep C630.

The set of substeps C622 to C632 is iterated for each of the sub-portions CU-i, CU ₂ , CU _k , ..., CU _N of the first portion CUi current of the current block CTU _U , in the order lexicographic predetermined.

Detailed description of the decoding part

An embodiment of the invention will now be described, in which the decoding method according to the invention is used to decode a data signal representative of an image or a sequence of images that is capable of being decoded. by a decoder according to any one of the present or future video decoding standards.

In this embodiment, the decoding method according to the invention is for example implemented in a software or hardware way by modifications of such a decoder. The decoding method according to the invention is represented in the form of an algorithm comprising steps D1 to D7 as represented in FIG. 8.

According to the embodiment of the invention, the decoding method according to the invention is implemented in a decoding device or decoder DO represented in FIG. 9.

As illustrated in FIG. 9, according to this embodiment of the invention, the decoder DO comprises a memory MEM_DO which itself comprises a buffer memory TAMP_DO, a processing unit UT_DO equipped for example with a microprocessor μΡ and driven by a PG_DO computer program that implements the decoding method according to the invention. At initialization, the code instructions of the computer program PG_DO are for example loaded into a RAM memory before being executed by the processor of the processing unit UT_DO.

The decoding method shown in FIG. 8 applies to a data signal representing a current image IC _j to be decoded or to a sequence of images to be decoded. For this purpose, information representative of the current image IC _j to be decoded is identified in the data signal F received at the decoder DO, as delivered following the coding method of FIG. 3.

With reference to FIG. 8, during a step D1, the quantized blocks CTUq-1, CTUq ₂ , ..., CTUq _u , CTUqs (1≤u≤S) are identified in the signal F. ) respectively associated CTU-i blocks, CTU ₂ , CTU _U , ..., CTUs previously coded according to the aforementioned lexicographic order, according to the coding method of Figure 3.

Such an identification step is implemented by a processor or MI_DO flow analysis software module, as shown in FIG. 9, said module being driven by the μΡ microprocessor of the processing unit UT_DO.

Other types of course than the one just described above are of course possible and depend on the order of course chosen coding.

Preferably, each of the blocks to be decoded CTU-i, CTU ₂ ,

CTU _U , ..., CTUs has a square shape and includes NxN pixels, with N> 2.

According to an alternative, each of the blocks to be decoded CTU-i, CTU ₂ , CTU _U , ..., CTUs has a rectangular shape and comprises NxP pixels, with N> 1 and P> 2.

During a step D2 shown in FIG. 8, the decoder DO of FIG. 9 selects as the current block the first quantized block CTUq _u which contains quantized data which has been coded during step C6 of FIG. .

During a step D3 represented in FIG. 8, in association with the quantized block CTUq _u that has been selected, the compressed value of the syntax element type_decoupe which has been selected at the reading is read. from step C4 of Figure 3 and, if applicable, the compressed value of the syntax element arr_decoupe1 or arr_decoupe2 associated with it.

As explained earlier in the description, the syntax element type_decoup denotes the indicator representative of a given subdivision mode. According to a preferred embodiment, the syntax element type_decoupe takes for example three values: - 0 to indicate a conventional subdivision of the current block into four rectangular or square blocks,

- 3 to indicate a lack of subdivision of the current block. The reading step D3 is performed by a processor or read software module ML_DO, as shown in FIG. 9, which module is driven by the microprocessor μΡ of the processing unit UT_DO.

In the same way as the coder CO of FIG. 4, are stored in the buffer memory TAMP_DO of the encoder DO of FIG. 9:

- a LT predefined _list of plural sets of coordinates each defining a rectangular block of a predetermined shape,

- a predefined LT _b list of several sets of coordinates each defining a rectangular block of a predetermined shape,

- a TC correspondence table.

During a step D4 shown in FIG. 8, the value of the cut-type syntax element which has been read in the aforementioned step D3 and, if applicable, the decoding of the value of the syntax element arr_decoupe1 or arr_decoupe2 associated with it.

The aforesaid step D4 is implemented by a processor or MDI indicator decoder software module as shown in FIG. 9, which module is driven by the microprocessor μΡ of the processing unit UT_DO.

During a step D5 shown in FIG. 8, the CTU _U current block is subdivided into at least a first portion CU 1 and a second portion CU ₂ , the first and second portions being complementary to one another. 'other. According to the invention:

the first part CUi has a rectangular or square shape, and the second part CU ₂ has a geometric shape with m sides, with m> 4.

According to a preferred embodiment, the current block CTU _U is subdivided: in a first part CUi of rectangular or square shape or in a plurality of parts of rectangular or square shape,

and at most a second portion CU ₂ of geometric shape with m sides.

Examples of subdivision have been presented with reference to the figures

5A and 5B above and will not be described again here.

The subdivision step D5 is performed by a processor or partitioning software module MP_DO, as shown in FIG. 9, which module is driven by the microprocessor μΡ of the processing unit UT_DO.

During a step D6 shown in FIG. 8, the CUi and CU ₂ portions of the CTU _U current block are decoded to be decoded according to a predetermined travel order. According to a preferred embodiment, the first part CUi is decoded before the second part CU ₂ . As an alternative, the first part CUi is decoded after the second part CU ₂ .

The decoding step D6 is implemented by a processor or decoder software module UDO as shown in FIG. 9, which module is driven by the microprocessor μΡ of the processing unit UT_DO.

As shown in more detail in FIG. 9, the UDO decoding module conventionally comprises:

a processor or software module MDE_DO of entropy decoding, for example of CABAC type ("Context Adaptive Binary Arithmetic Coder" in English) or a Huffman decoder known as such,

a processor or software module MQ1 ^"1 _DO dequantization,

a processor or software module MT1 ^"1 _DO of inverse transformation of DCT ¹ type (abbreviation of" Discrete Cosine Transform "),

DST ¹ (abbreviation of "Discrete Sine Transform"), DWT ¹ (abbreviation of "Discrete Wavelet Transform"),

a processor or software module PRED1 ^"1 _DO of inverse prediction,

a processor or module CAL2_DO of block reconstruction calculation.

At the end of step D6, a decoded current block CTUD _U is obtained. During a step D7 shown in FIG. 8, said decoded block CTUD _U is written in a decoded image ID _j .

Such a step is implemented by a processor or software module URI image reconstruction as shown in Figure 9, said module being controlled by the microprocessor μΡ processing module UT_DO.

The decoding steps that have just been described above are implemented for all the CTU-i, CTU ₂ , CTU _U , ..., CTU _S blocks to be decoded from the current image IC _j considered, in a predetermined order which is for example the lexicographic order.

Other types of course than the one just described above are of course possible.

A first embodiment of the different substeps implemented during the aforementioned decoding step D6 in the UDO decoding module shown in FIG. 9 will now be described with reference to FIG. 10A.

According to this first embodiment, the data signal F contains the partitioning indicators of a CTU current block _U which has been coded according to the embodiment of FIG. 6A. For this purpose, as described above in the description in conjunction with the embodiment of FIG. 6A, the signal F contains the following four values 1 133 which have been decoded at the end of the aforementioned step D4 and which are representative:

partitioning of the current block CTU _U according to one of the subdivision modes represented in FIG. 5A and more precisely according to the subdivision type SUBD2i of FIG. 5A,

and the absence of subdivision of the CUi and CU _{2 parts} of the CTU _U current block.

As a variant, the data signal F contains the following three values 1 13, in the case where the indicator type_cutter of value 3 associated with the coded data of the second part CU ₂ has not been written in the data signal F, considering that the second portion CU ₂ defines a homogeneous area of the current block CTU _U. Consequently, the cut_type flag is systematically set to the predetermined value 3, so that the second portion CU ₂ is not subdivided at decoding.

During a sub-step D610 shown in FIG. 10A, the decoding module UDO selects a set of current quantized coefficients CUq _k associated with the current portion CU _k (k = 1 or k = 2), first of all the set of quantized coefficients associated with the square portion CU-i, first of all the set of quantized coefficients associated with the part CU ₂ with m sides.

During a substep D61 1 shown in FIG. 10A, an entropic decoding of the set of quantized coefficients CUqi current associated with the first part CU-i is carried out. In the preferred embodiment, the decoding performed is an entropy decoding of arithmetic type or Huffman. The substep D61 1 then consists of:

reading the symbol or symbols of a predetermined set of symbols which are associated with the set of quantized Cuqi coefficients current,

- Associating digital information, such as bits, to the symbol (s) read (s).

At the end of the substep D61 1 above, is obtained a plurality of digital information associated with the set of quantized coefficients Cuqi current.

Such a substep D61 1 entropy decoding is implemented by the entropy decoding module MDE_DO shown in Figure 9.

During the aforementioned sub-step D61 1, the information relating to the predictive coding of the portion CUi as implemented in the substep C61 1 of FIG. 6A, which have been written in the signal of FIG. F. Such reconstruction information includes in particular the type of prediction (inter or intra), and if appropriate, the intra prediction mode or the reference image index and the displacement vector used in the mode of the prediction. inter prediction

During a substep D612 shown in FIG. 10A, dequantization of the digital information obtained following substep D61 1 is carried out according to a conventional dequantization operation which is the inverse operation of the quantification implemented during the substep Quantization C614 of Figure 6A. A set of dequantized coefficients CUDqi current is then obtained at the end of substep D612. Such a sub-step D612 is performed by means of the dequantization module MQ ^"1 _DO, as represented in FIG. 9.

During a substep D613 shown in FIG. 10A, the current set of dequantized coefficients CUDqi is transformed, such a transformation being an inverse direct transformation, such as, for example, a discrete inverse cosine transformation. type DCT ^{^"1.} This transformation is the inverse of the transformation performed to the C613 substep of Figure 6A. at the end of the D613 substep is obtained residue decoded part CUDr-,. a this operation is performed by the MT module ^"1 _DO shown in FIG. 9.

During a sub-step D614 represented in FIG. 10A, the PRED module ^"1 _DO" of FIG. 9 performs the predictive decoding of the current portion CUi with the aid of the information relating to the predictive coding of the CUi portion which has were decoded during the substep D61 1 above.

Said aforementioned predictive decoding sub-step makes it possible to construct a predicted part CUDpi which is an approximation of the current part CUi to be decoded.

During a substep D615 shown in FIG. 10A, the module

CAL2_DO of FIG. 9 proceeds to the reconstruction of the current part CUi by adding to the decoded residue part CUDn, obtained at the end of the substep D613, the predicted part CUDpi which was obtained at the end of the sub-step step D614 above.

The aforementioned substeps D610 to D615 are then iterated in order to decode the second part CU ₂ with m sides of the current block CTU _U.

According to the invention, in the case of decoding the second portion CU ₂ with m sides, one or more reconstruction information pixels of the second portion CU ₂ are set to predetermined values.

Thus, preferably, during the sub-step D614 for predictive decoding of the portion CU ₂ of the current block CTU _U , the pixels of the portion CU ₂ to be decoded are predicted with respect to pixels of predetermined corresponding values, respectively. Such values are stored in a list LP contained in the buffer memory TAMP_DO of the decoder DO of FIG. 9.

According to a preferred embodiment, the sub-step D610 of FIG. 10A is not implemented since no set of quantized coefficients associated with the portion CU ₂ with m sides has been transmitted in the data signal F The quantized coefficients of the quantized residual portion CUq ₂ are then all directly set to zero by the UDO decoding module of FIG.

Such an arrangement is made advantageous by the fact that the CU ₂ part of the CTU _U current block which has been coded is considered homogeneous.

According to another preferred embodiment, the aforementioned sub-step D61 1 is not implemented in its entirety, the decoder DO deducing directly, following the abovementioned substep D610, predetermined values of associated reconstruction information. at the residue portion CUr ₂ .

As a variant, the pixels of the portion CU ₂ to be decoded are predicted conventionally, in the same way as the portion CU-i.

According to the invention, between the substeps D61 1 and D612 mentioned above, an intermediate step D61 is implemented. During this intermediate step, the decoded pixel values that have been obtained as a result of the entropy decoding step of the plurality of digital information associated with the current set of quantized coefficients CUq ₂ are supplemented by values of predetermined pixels, until a block of square or rectangular pixel values is obtained.

According to various possible embodiments, the pixel values associated with the set of quantized coefficients CUq ₂ current can be completed:

by respective values of null pixels,

by pixel values reconstructed conventionally by interpolation, by pixel values reconstructed conventionally using the so-called "inpaiting" technique.

The substep D61 10 mentioned above is implemented by a calculation software module CAL1_DO as shown in FIG. 9, which module is driven by the microprocessor μΡ of the processing unit UT_DO.

Such an arrangement makes it possible to reuse the transformation software module MT ^"1 _DO of FIG. 9, which conventionally applies transformations of square or rectangular blocks.

In view of the fact that sub-step D61 applies only to the decoded pixel values that were obtained as a result of the entropy decoding step of the plurality of digital information associated with the set of coefficients quantized CUq ₂ geometric-shaped current with m sides, this step, as well as the calculation module CAL1_DO, are represented in dashed lines, respectively in FIGS. 10A and 9.

A second embodiment of the different substeps implemented during the aforementioned decoding step D6 in the encoding module UDO shown in FIG. 9 will now be described with reference to FIG. 10B.

This second embodiment differs from that of FIG. 10A in that the first portion CU i to be decoded from the CTU current block _U is subdivided again.

According to this second embodiment, the data signal F contains the partitioning indicators of a CTU current block _U which has been coded according to the embodiment of FIG. 6B. For this purpose, as described above in the description in conjunction with the embodiment of FIG. 6B, the signal F contains the following fifteen values as shown in FIG. 7 and which have been decoded at the end of FIG. step D4 above.

Such values are representative:

partitioning of the CTU _U current block according to one of the subdivision modes represented in FIG. 5A and more precisely according to the subdivision type SUBD2 of FIG. 5A, the absence of subdivision of the second part CU ₂ with m sides of the current block CTU _U ,

the subdivision of the first part CUi of the current block CTU _U as represented in FIG. 7.

As a variant, the data signal F does not contain the two values equal to 3 represented in bold, in the case where:

the indicator type_cut of value 3 associated with the coded data of the second part CU ₂ has not been entered in the data signal F, taking into account the fact that the second part CU ₂ defines a homogeneous zone of the current block CTU _U ,

the indicator type_cutting value 3 associated with the coded data of the second part CU22i with m sides of the block CU2i as represented in FIG. 7, given that the second part CU22 defines a homogeneous zone of the block CU2-I.

Consequently, the cut_type indicator is systematically set to the predetermined value 3, so that neither the second portion CU ₂ of the current block CTU _U to be decoded, nor the second portion CU22 m-side of the block CU2 of the current block CTU _U to decode, is subdivided at decoding.

During a sub-step D620 represented in FIG. 10B, the decoding module UDO selects a set of quantized coefficients CUq _k current associated with the current part CU _k (k = 1 or k = 2), first of all the set of quantized coefficients associated with the square portion CU-i, first of all the set of quantized coefficients associated with the part CU ₂ with m sides.

During a substep D621 shown in FIG. 10B, the decoding module UDO tests whether the index k associated with the current part CU _k to be decoded is 1 or 2.

If the index k is equal to 2, the portion CU ₂ of the current block CTU _U to be decoded is decoded according to the substeps D610 to D615 of FIG. 10A.

If the index k is equal to 1, during a substep D622 shown in FIG. 10B, the decoding module UDO of FIG. 9 selects a current subpart CU to be decoded from the first portion CUi of the block current CTU _U to be decoded, such that 1 <k'≤N. In the example represented in FIG. 7, N = 8, since the first portion CUi of the current block CTU _U has been subdivided into eight sub-parts of the "coding unit" type CU1 _1; CU21 _1; CU22 _1; CU31 _1; CU32 _1; CU33 _1; CU34 _1; CU4L

During a substep D623 shown in FIG. 10B, the entropy decoding module MDE_DO of FIG. 9 performs an entropy decoding of the set of quantized coefficients CUq _k 'current associated with the current subpart CU _k of the first part CUi of the current block CTU _U to be decoded. In the preferred embodiment, the decoding performed is an entropy decoding of arithmetic type or Huffman. Sub-step D623 then consists of:

reading the symbol (s) of a predetermined set of symbols which are associated with the set of quantized coefficients Cllq ^ current,

- Associating digital information, such as bits, to the symbol (s) read (s).

At the end of the substep D623 above, is obtained a plurality of digital information associated with the set of quantized coefficients CUqk 'current.

During the sub-step D623, the entropy decoding module MDE_DO of FIG. 9 also performs an entropy decoding of the indicator _PR representative of the inter or intra prediction mode that has been selected for this current subpart CU during sub-step C623 of Figure 6B.

During an optional sub-step D624 shown in FIG. 10B, in the case where the current sub-portion CU _k > to be decoded has been subdivided during the substep C626 of FIG. 6B into a plurality W of prediction sub-parts PU-i, PU ₂ , PU _z , ... PUw (1≤z≤W), the read-only software module ML_DO of FIG. 9 proceeds to read the compressed value of the representative indicator such subdivision. Such an indicator consists of the syntax element type_cut and, if applicable, the syntax element arr_cut1 or cut_decoup2 that is associated with it.

During an optional substep D625 shown in FIG. 10B, the MDI indicator decoding software module of FIG. 9 proceeds to decode the value of the syntax element type_decut that has been read from FIG. substep D624 above and, if applicable, the decoding of the value of the syntax element arr_decoupe1 or arr_decoupe2 associated with it.

During an optional sub-step D626 shown in FIG. 10B, the partitioning software module MP_DO of FIG. 9 sub-divides the current sub-part CU 1 to be decoded into a plurality W of prediction subparts PU ₁ , PU ₂ , PU _Z , ..., PU _W (1 <z≤W).

During an optional sub-step D627 shown in FIG. 10B, the decoding module UDO of FIG. 9 selects a first current sub-portion PU _Z. Such a selection is made in a predefined order, such as for example the lexicographic order.

During an optional substep D628 shown in FIG. 6B, the entropy decoding module MDE_DO of FIG. 9 proceeds, in association with the current subpart PU _Z , to an entropy decoding of the optimal prediction parameters which have were selected during substep C626 of FIG. 6B, in association with the _PR indicator, which is representative of the prediction mode selected in substep C623 above and decoded in substep D623. If, for example, the prediction mode INTER has been selected in the aforementioned sub-step C623, the decoded optimal prediction parameters are one or more motion vectors, as well as one or more reference images. If, for example, the INTRA prediction mode has been selected in the aforementioned sub-step C623, the optimal prediction parameters are associated with an INTRA mode selected from among various available INTRA modes.

The substeps D627 to D628 are iterated for each of the sub-parts PU-i, PU ₂ , PU _Z ,..., PU _W of the current subpart CU _k ^> to be decoded from the first part CU i of the current block CTU _U , in the predetermined lexicographic order.

During an optional sub-step D629 shown in FIG. 10B, in the case where the current sub-portion CU _k > to be decoded has been subdivided, during the sub-step C627 of FIG. 6B, into a plurality Z of transform subparts TU-i, TU ₂ , TU _w , ... TUz (1≤w≤Z), the read software module ML_DO of FIG. 9 proceeds to read the compressed value of the representative indicator of such subdivision. Such an indicator consists of the cut_type syntax element and, if applicable, the syntax element arr_decoupe1 or arr_decoupe2 associated with it.

In an optional substep D630 shown in FIG. 10B, the MDI indicator decoding software module of FIG. 9 decodes the value of the syntax element type_cut which has been read to the subscript. said step D629 and, if applicable, decoding the value of the syntax element arr_decoupe1 or arr_decoupe2 associated with it.

During an optional sub-step D631 shown in FIG. 10B, the partitioning software module MP_DO of FIG. 9 sub-divides the current sub-part CU _k 'to be decoded into a plurality Z of sub-parts of transform TU ₁ . TU ₂ , ..., TU _W , ..., TU _Z (1 <w _{Z Z} ).

During an optional sub-step D632 shown in FIG. 6B, the decoding module UDO of FIG. 9 selects the set of quantized coefficients TUq _w current associated with the first current transform sub-part TU _W. Such a selection is made in a predefined order, such as for example the lexicographic order.

During a sub-step D633 represented in FIG. 6B, the entropy decoding module MDE_DO of FIG. 9 performs an entropy decoding of the set of quantized coefficients TUq _w current associated with the first current transform sub-part. TU _W to decode. In the preferred embodiment, the decoding performed is an entropy decoding of arithmetic type or Huffman. Sub-step D633 then consists of:

- Associating digital information, such as bits, to the symbol (s) read (s).

At the end of the substep D633 above, is obtained a plurality of digital information associated with the set of quantized coefficients TUq _w current.

During a substep D634 shown in FIG. 10B, the dequantization module MQ ^"1 _DO of FIG. 9 dequantizes the digital information obtained as a result of substep D633, according to a conventional operation of FIG. dequantization which is the reverse operation of the quantization implemented during the quantization sub-step C631 of FIG. 6B. A set of dequantized coefficients TUDq _w current is then obtained at the end of substep D634.

During a substep D635 shown in FIG. 10B, the MT module ^"1 _DO of FIG. 9 carries out a transformation of the set of dequantized coefficients TUDq _w current, such a transformation being a direct inverse transformation, such as that, for example, a reverse discrete cosine transformation of DCT type ^1. This transformation is the inverse operation of the transformation carried out in substep C630 of FIG.6A.After completion of substep D635, there is obtained a decoded residue part TUDr _w .

During a sub-step D636 shown in FIG. 10B, the PRED module ^"1 _DO" of FIG. 9 proceeds to the predictive decoding of the first current transform subunit TU _W using the optimal prediction parameters which were read during the aforementioned substep D628.

Said aforementioned predictive decoding sub-step makes it possible to construct a first predicted current transform subunit TUDp _w which is an approximation of the first current transform sub-part TU _W to be decoded.

During a substep D637 shown in FIG. 10B, the module CAL2_DO of FIG. 9 proceeds to the reconstruction of the first sub-part of the current transform TU _W by adding to the decoded residue part TUDr _w , obtained at From the substep D635, the predicted part TUDp _w which was obtained at the end of the substep D636 above.

The set of substeps D632 to D637 is iterated for each of the subsections TU-i, TU ₂ , TU _W ,..., TU _Z to be decoded of the sub-part CU _k ^> current to be decoded from the first CUi part of CTU current block _U , in the predetermined lexicographic order.

According to the invention, in the case where the current transformed sub-part TU _W has a geometric shape with m sides, an intermediate substep D6330 is implemented between substeps D633 and D634 mentioned above. During this intermediate substep, the decoded pixel values that were obtained as a result of the entropy decoding sub-step D633 of the plurality of digital information associated with the set of quantized coefficients TUq _w current are supplemented by predetermined pixel values until a block of square or rectangular pixel values is obtained.

The substep D6330 mentioned above is implemented by the calculation software module CAL1_DO as represented in FIG. 9.

The set of substeps D622 to D637 is iterated for each of the sub-portions CU-i, CU ₂ , CU _k , ..., CU _N to be decoded from the first part CUi current of the current block CTU _U , in the predetermined lexicographic order.

It goes without saying that the embodiments which have been described above have been given for purely indicative and non-limiting purposes, and that many modifications can easily be made by those skilled in the art without departing from the scope. of the invention.

Claims

1. A method of encoding at least one image (IC _j ), comprising a step of subdividing the image into a plurality of blocks (CTU-i, CTU ₂ , CTU ,, ..., CTUs),

characterized in that it comprises the following steps:

- subdivide (C2) at least one current block into a first (CU-i) and a second (CU ₂ ) part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the block current, said second part having a geometric shape with m sides, with m> 4,

coding (C6) the first and second parts.

2. Encoding device (CO) for at least one image (IC _j ), comprising partitioning means for subdividing the image into a plurality of blocks (CTUi, CTU _2> .... CTUi, ..., CTUs)

characterized in that the partitioning means (MP_CO) are able to subdivide at least one current block into a first and a second part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the current block, said second part having a geometric shape with m sides, with m> 4,

and in that it comprises coding means (UCO) for coding the first and second parts.

A computer program comprising instructions for implementing the encoding method according to claim 1, when executed on a computer.

4. A method of decoding a data signal (F) representative of at least one coded picture having been subdivided into a plurality of blocks (CTU-i,

characterized in that it comprises the following steps: - subdivide (D5) at least one current block into a first (CU-i) and a second (CU ₂ ) part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the block current, said second part having a geometric shape with m sides, with m> 4,

- decode (D6) the first and second parts.

5. decoding method according to claim 4, wherein during the decoding step of the second part m-sides of the current block, at least one pixel reconstruction information of said second part m-sides is set to a predetermined value.

6. Decoding method according to claim 5, wherein said at least one pixel reconstruction information of the second part m-sides of the current block is representative of the absence of subdivision of said second part m-sides of the current block.

7. decoding method according to claim 5, wherein said at least one pixel reconstruction information of the second part m-sides of the current block is representative of the absence of residual information resulting from a prediction of the pixels of said second part with m sides of the current block.

8. decoding method according to claim 5, wherein said at least one pixel reconstruction information of the second part m-sides of the current block is representative of predetermined prediction values of the pixels of said second part m-sides of the current block .

9. Decoding method according to any one of claims 4 to 8, comprising, prior to the step of subdivision of the current block, a reading step (D3), in the data signal, information indicating whether the current block is intended to be subdivided according to claim 4 or according to another predetermined method.

10. Decoding method according to any one of claims 4 to

9, comprising, prior to the step of partitioning the current block, a step of reading (D3), in the data signal, information indicating a subdivision configuration of the current block selected from among different predetermined subdivision patterns.

1 1. Decoding method according to any one of claims 4 to

10, wherein the step of decoding the second part of the current block with m sides comprises the substeps consisting of:

applying (D61 1) an entropy decoding to the pixels of said second part with m sides,

- Complete (D61 10) entropically decoded pixels of said second part m-sides by pixels reconstructed according to a predetermined reconstruction method, until a block of square or rectangular pixels.

12. Decoding method according to any one of claims 4 to

1 1, wherein a subdivided current block contains at most a portion having a geometric shape with m sides.

13. Device for decoding a data signal (F) representative of at least one encoded image having been subdivided into a plurality of blocks (CTUi, CTU _2, ..., CTUi, ..., _s CTU )

characterized in that it comprises:

- partitioning means (MP_DO) for subdividing at least one current block into a first (CU-i) and a second (CU ₂ ) part, the first part having a rectangular or square shape and the second part forming the complement of the first part in the current block, said second part having a geometric shape with m sides, with m> 4,

means (MDE_DO) decoding for decoding the first and second parts.

A computer program comprising instructions for implementing the decoding method according to any one of claims 4 to 12, when executed on a computer.