JP6130515B2 - High throughput residual coding for transform skip blocks for CABAC in HEVC - Google Patents

Description

CROSS REFERENCE TO RELATED APPLICATIONS This application is the United States of America entitled "HIGH THROUGHPUT BINARIZATION (HTB) METHOD FOR CABAC IN HEVC", filed on January 19, 2012, in HEVC. US patent entitled “LOSSLES CODING TECHNIQUE FOR CABAC IN HEVC” filed on Jan. 27, 2012, which is a continuation-in-part of patent application No. 13 / 354,272 "High THROUGHP High-Throughness Map Processing for CABAC in HEVC", filed February 2, 2012, which is a continuation-in-part of application 13 / 360,615. “A different parameter selection technique for CABAC in HEVC” filed on April 11, 2012, which is a continuation-in-part of US patent application No. 13 / 365,215 entitled “T SIGNIFICANCE MAP PROCESSING FOR CABAC IN HEVC”. "HEVC" filed on April 26, 2012, which is a continuation-in-part of US Patent Application No. 13 / 444,710 entitled "LOSSLESS CODING WITH DIFFERENT PARAMETER SELECTION TECHNIQUE FOR CABAC IN HEVC" US Patent entitled “High THROUGHPUT CODING FOR CABAC IN HEVC” No. 13 / 457,272, which is a continuation-in-part of US Patent Application No. 13 / 857,366, each of which is hereby incorporated by reference in its entirety. Incorporated by reference.

TECHNICAL FIELD The present disclosure relates generally to electronic devices. More specifically, the present disclosure relates to an electronic device that uses extended context adaptive binary arithmetic coding (CABAC) for encoding and / or decoding.

  Many decoders (and encoders) receive (provide in the case of an encoder) encoded data for a block of images. Typically, an image is divided into a plurality of blocks, and each block is encoded in some way, for example using a discrete cosine transform (DCT), and supplied to a decoder. The block may indicate a rectangular area in the image or may be composed of pixels. For example, a 16 × 16 block is an area having a width of 16 pixels and a height of 16 pixels. The decoder receives the encoded blocks and decodes each block in some way, such as using an inverse discrete cosine transform.

  Video coding standards such as MPEG-4 part 10 (H.264) compress video data for transmission over channels with limited frequency bandwidth and / or limited storage capacity. These video coding standards include, for example, intra prediction, spatial domain to frequency domain transformation, quantization, entropy coding, motion estimation, and motion compensation to more effectively encode and decode frames. It includes a plurality of encoding stages.

  International Telecommunication Union Telecommunication Standardization Sector (ITU-T) Study Group 16 (Study Group 16: SG16) Working Group 3 (International Working Party 3: WP3) (International Organization for Standardization / International Electrotechnical Commission: ISO / IEC) Joint Technical Committee 1 / Subcommittee 29 / Work Group 11 (Joint Technical Committee 1 / Subcommitte) 9 / Working Group 11: JTC1 / SC29 / WG11), a joint work team on video coding (JCT-VC) A standardization effort on the video coding standard called Like some prior video coding standards, HEVC is block-based coding. An example of a known HEVC encoder is shown in FIG. HEVC decoders are also known.

  In HEVC, context adaptive binary arithmetic coding (CABAC) is used to compress transformed and quantized coefficients (TQC) without loss. TQC is defined in an encoder by processing an image block with forward transform to generate transform coefficients, and then quantizing the transform coefficients using an operation that maps a plurality of transform coefficient values to TQC values. The TQC values are then communicated to the decoder as coefficient level values or level values, and the level values for each coefficient are then mapped to transform coefficient values that are similar but not necessarily the same as the transform coefficient values calculated at the encoder. Is done. CABAC-based encoding and / or decoding techniques are generally context-adaptive, which is (i) adaptive to symbols based on the value of previous symbols previously encoded and / or decoded. And (ii) Context identifying the set of previously encoded and / or decoded symbols used for adaptation. Past symbols may be located in spatially and / or temporally adjacent blocks. In many cases, the context is based on the symbol values of neighboring blocks.

  As mentioned above, CABAC can be used to compress TQC without loss. As background, the TQC may be derived from different block sizes according to the transform size (eg, 4 × 4, 8 × 8, 16 × 16, 32 × 32, 16 × 32). Prior to entropy coding, a two-dimensional (2D) two-dimensional (TQC) may be converted to a one-dimensional (1D) array. In one example, the 2D array TQC in the 4 × 4 block may be arranged as shown in Table (1).

  When converting 2D TQC to a 1D array, blocks may be scanned diagonally in a zigzag pattern. Continuing with this example, the 2D array TQC shown in Table (1) has the first row, first column, the first row, second column, the second row, first column, the third row, first column, the second row, second column. By scanning 2 columns, 1st row 3rd column, 1st row 4th column, 2nd row 3rd column, 3rd row 2nd column, 4th row 1st column, and so on, 1D array TQC [ 4,0,3, -3,2,1,0, -1,0,. . . ] May be converted to.

  A 1D sequence of TQC is represented by a string of syntax elements (SE) in CABAC. An example SE column for an example of a TQC 1D array is shown in FIG. SE represents the following parameter. That is, Last_position_X / Y, significance map, attributes Greator_than_1, Greeter_than_2, positive / negative information, and Absolute-3. Last_position_X / Y represents the position (X / Y) of the final non-zero coefficient in the corresponding block. The significance map represents the significance of each coefficient. Greeter_than_1 indicates whether the coefficient amplitude is greater than 1 for each non-zero coefficient (ie, a coefficient having a significance flag of 1). Greater_than_2 indicates, for each coefficient having an amplitude greater than 1 (ie, having a Greeter_than_1 flag of 1), that coefficient amplitude is greater than 2.

  In CABAC in HEVC, a typical SE is encoded. FIG. 3 shows the CABAC framework used to encode SE. The CABAC encoding technique includes encoding symbols using multiple stages. In the first stage, CABAC uses a “binarizer” to map input symbols to a sequence of binary symbols or “bins”. The input symbols may be non-binary symbols that are binarized before being encoded into bits, or otherwise transformed into a sequence of binary (1 or 0) symbols. Bins may be encoded into bits using either a “bypass encoding engine” or a “regular encoding engine”.

  For the regular coding engine in CABAC, a probability model is selected in the second stage. A probabilistic model is used to arithmetically encode one or more bins of a binarized input symbol. This model may be selected from a list of available probability models, depending on the context that is a function of the recently encoded symbols. The probability model stores the probability that the bin is “1” or “0”. In the third stage, the arithmetic encoder encodes each bin according to the selected probability model. For each bin, there are two subranges corresponding to “0” and “1”. The fourth stage involves updating the probability model. The selected probability model is updated based on the actual encoded bin value (eg, if the bin value is “1”, the frequency number of “1” is increased). The decoding technique for CABAC decoding reverses this process.

  For the CABAC bypass coding engine, the second stage involves converting bins into bits, omitting the computationally expensive context estimation and probability update stages. The bypass encoding engine assumes a fixed probability distribution for the input bin. The decoding technique for CABAC decoding reverses this process.

  CABAC conceptually encodes symbols using two steps. In the first step, CABAC performs binarization from input symbols to bins. In the second step, CABAC performs bin to bit conversion using either a bypass coding engine or a regular coding engine. The resulting encoded bit value is supplied to the decoder in the bitstream.

  CABAC conceptually decodes symbols using two steps. In the first step, CABAC converts input bits to bin values using either a bypass decoding engine or a regular decoding engine. In the second step, CABAC performs de-binarization to recover the transmitted symbol value for the bin value. The recovered symbol may be non-binary in nature. In the remaining aspects of the decoder, the recovered symbol values are used.

  As described above, the CABAC encoding and / or decoding process includes at least two different operation modes. In the first mode, the probability model is updated based on the actually encoded bin values, and this mode is commonly referred to as the “regular encoding mode”. Regular coding mode requires several consecutive serial operations and the associated calculations are complex and take a considerable amount of time to complete. In the second mode, the probability model is not updated based on the actually encoded bin values, and this mode is generally referred to as the “bypass encoding mode”. In the second mode, there is no probability model for decoding bins (possibly other than a fixed probability), so there is no need to update the probability model.

  When using CABAC encoding in HEVC, it depends on different factors such as, but not limited to, the total number of bins / pixels, the number of bypass bins / pixels, and the number of regular (or context) encoding bins / pixels. Throughput performance can vary. Generally speaking, the throughput in the case of high bit rate coding (low quantization parameter (QP) value) is significantly lower than the throughput in other cases. Thus, throughput at high bit rates can consume a significant amount of processing resources and / or can take a significant amount of time for encoding / decoding. The present disclosure below addresses this and other problems.

  In addition, it is known that CABAC can be used in a lossless coding mode for compressing residual samples. In one example, the residual sample is a value corresponding to a particular position in the image. Typically, the residual sample corresponds to the difference between a value corresponding to a specific position in the image and a predicted value corresponding to the same specific position in the image. Alternatively, the residual sample is a value corresponding to a particular position in the image that has not been processed by the transformation operation or that uses a transformation operation that is not normally used to create a TQC. The residual samples can come from different block sizes according to their sample size (4 × 4, 8 × 8, 16 × 16, 32 × 32, 16 × 32, etc.). Similar to TQC coding, before entropy coding, the 2D residual sample block is first converted to a 1D array. In one example, 2D array residual samples in a 4 × 4 block may be arranged as shown in Table (2).

  When converting a 2D residual sample to a 1D array, the blocks may be scanned diagonally in a zigzag pattern. Continuing with this example, the 2D array residual samples shown in Table (2) are: first row, first column, first row, second column, second row, first column, third row, first column, second 1D array by scanning row 2nd column, 1st row 3rd column, 1st row 4th column, 2nd row 3rd column, 3rd row 2nd column, 4th row 1st column, and so on Residual samples [4,0,3, -3,2,1,0, -1,0,. . . ] May be converted.

  A 1D array of residual samples is represented in CABAC by a sequence of syntax elements (SE). FIG. 11 shows an example of an SE column for a 1D array example of residual samples. SE represents the following parameter. That is, Last_position_X / Y, significance map, attributes Greater_than_1, Greeter_than_2, positive / negative information, and Absolute-3.

  In the CABAC lossless coding mode in HEVC, a typical SE is coded. To encode the SE, the CABAC framework of FIG. 3 may be used. The CABAC encoding technique includes encoding symbols using multiple stages. In the first stage, CABAC uses “binarizers” to map input symbols to binary symbols, or “bin” columns. The input symbols may be non-binary symbols that are binarized before being encoded into bits, or otherwise transformed into a sequence of binary (1 or 0) symbols. Bins may be encoded into bits using the “regular encoding engine” described above.

  For a regular coding engine in the CABAC lossless coding mode, a probabilistic model (also known as a “context model” in the CABAC lossless coding mode) is selected in the second stage. This model is used to arithmetically encode one or more bins of a binarized input symbol. This model may be selected from a list of available models depending on the context that is a function of the recently encoded symbols. This model stores the probability that the bin is “1” or “0”. In the third stage, the arithmetic encoder encodes each bin according to the selected model. For each bin, there are two subranges corresponding to “0” and “1”. The fourth stage involves updating the model. The selected model is updated based on the actual encoded bin value (eg, if the bin value is “1”, the frequency number of “1” is increased). The decoding technique for CABAC decoding reverses this process.

  The number of models used as described in the previous paragraph may be 184. Specifically, 36 models are used for Last_position_X / Y (18 models for Last_position_X, 18 models for Last_position_Y), and 48 models are used for the significance map (4 × 4 blocks: luminance 9, color difference 6 8 × 8 block: luminance 11, color difference 11; 16 × 16 or 32 × 32 block: luminance 7, color difference 4), 100 models are used for attributes Greator_than_1, Greeter_tan_2, positive / negative information, and Absolute-3 (luminance Greater_than_1) Flag: 30; Color difference Greeter_than_1 flag 20; Luminance Greeter_than_2 flag: 30; and Color difference Greeter_than_2 flag 20).

  When using CABAC coding in HEVC in lossless coding mode, encoding / decoding is computationally complex. One reason for this computational complexity is to use 184 models, as explained above. Because of this computational complexity, encoding / decoding can consume a significant amount of processing resources and / or can take a significant amount of time to complete. The present disclosure below addresses this and other problems.

  One aspect of the present invention provides a system that includes a first electronic device of a decoder, the first electronic device obtaining a bitstream and recovering a binary symbol from the obtained bitstream. Responsive to determining whether the binary symbol is decoded using a high-throughput residual coding mode and determining that the binary symbol is not decoded using a high-throughput residual coding mode. To obtain a block of coefficients (TQC) transformed and quantized using the first coding technique and to determine that the binary symbols are decoded using the high-throughput residual coding mode. In response, obtaining a residual sample using a second different encoding technique, and obtaining a block of the obtained TQC or obtained residual sample, or obtained TQ. Composed video data representing blocks or resulting residue sample to perform and be stored in the memory device.

  Another aspect of the invention is the steps of obtaining a bitstream, recovering binary symbols from the obtained bitstream, and whether the binary symbols are decoded using a high-throughput residual coding mode. And transformed and quantized coefficients (TQC) using the first encoding technique in response to determining that the binary symbols are not decoded using the high-throughput residual coding mode And obtaining a residual sample using a second different coding technique in response to determining that the binary symbol is decoded using a high throughput residual coding mode; , A block representing the obtained TQC or the resulting residual sample, or a block representing the obtained TQC block or the obtained residual sample. The provides a A method comprising the steps of storing in a memory device Odeta.

It is a block diagram of a HEVC encoder. It is a table | surface which shows the row | line | column of the syntax element according to CABAC. FIG. 6 is a block diagram of the CABAC framework for a sequence of syntax elements. It is a block diagram which shows the example of an encoder and a decoder. 2 is a flowchart illustrating one configuration of a method for a high-throughput binarization mode in an electronic device. It is a flowchart which shows one structure of the encoder process using high-throughput binarization mode. 6 is a flowchart showing one configuration of a method for a high-throughput binarization mode in a decoding-side electronic device. It is a flowchart which shows one structure of the decoder process using high-throughput binarization mode. FIG. 6 shows a mapping table that can be used to define input values in high throughput binarization mode. It is a figure which shows the some binarization table | surface which can be used for the adaptive binarization in high throughput binarization mode. It is a table | surface which shows the string of the syntax element according to the lossless encoding mode in CABAC. It is a block diagram which shows the example of the encoder and decoder with respect to a lossless encoding technique. 2 is a flowchart illustrating one configuration of a method for lossless encoding in an electronic device. It is a table | surface which shows the column of the syntax element according to the structure shown by FIG. It is a flowchart which shows one structure of the method for the lossless decoding in the electronic device of a decoding side. 5 is a flowchart illustrating another configuration of a method for lossless encoding in an electronic device. It is a flowchart which shows another structure of the method for the lossless encoding in the electronic device of a decoding side. 12 is a flowchart illustrating yet another configuration of a method for lossless encoding in an electronic device. It is a flowchart which shows another structure of the method for the lossless encoding in the electronic device of a decoding side. 6 is a flowchart illustrating an example of a configuration of an encoder or a decoder for determining whether a high-throughput binarization mode condition is satisfied. 6 is a flowchart illustrating an example of a configuration of an encoder or a decoder for determining whether a high-throughput binarization mode condition is satisfied. 6 is a flowchart illustrating an example of a configuration of an encoder or a decoder for determining whether a high-throughput binarization mode condition is satisfied. 6 is a flowchart illustrating an example of a configuration of an encoder or a decoder for determining whether a high-throughput binarization mode condition is satisfied. 6 is a flowchart illustrating an example of a configuration of an encoder or a decoder for determining whether a high-throughput binarization mode condition is satisfied. 6 is a flow diagram illustrating one configuration of a method for determining whether a high throughput mode condition is satisfied in a decoding-side electronic device. 6 is a flow diagram illustrating another configuration of a method for determining whether a high throughput mode condition is satisfied in a decoding-side electronic device. It is a block diagram which shows the example of an encoder and a decoder. 6 is a flow diagram illustrating one configuration of a method for high-throughput significance map decoding in a decoding-side electronic device. 6 is a flow diagram illustrating another configuration of a method for high-throughput significance map decoding in a decoding-side electronic device. 6 is a flowchart illustrating one configuration of a method for high-throughput significance map decoding with a decoding bypass function in a decoding-side electronic device. 12 is a flowchart showing one configuration of a method for high-throughput significance map decoding with a decoding method switching function in a decoding-side electronic device. It is a table | surface used in order to update a rice parameter according to the lossless encoding mode in CABAC. It is a block diagram which shows the example of an encoder and a decoder. 3 is a flow diagram illustrating one configuration of a method for lossless encoding with different parameter selection in an electronic device. 6 is a flow diagram illustrating one configuration of a method for lossless encoding with different parameter selection in a decoding-side electronic device. It is a figure which shows the example of the syntax element produced | generated according to CABAC. It is a block diagram which shows the example of an encoder and a decoder. 2 is a flowchart illustrating one configuration of a method for high throughput encoding for CABAC in HEVC in an electronic device. 6 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in an electronic device on the decoding side. FIG. 35 is a diagram illustrating an example of syntax elements generated according to the configuration of FIG. 34. It is a figure which shows the rice parameter update table reduced. 6 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in an electronic device on the decoding side. 6 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in an electronic device on the decoding side. 6 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in an electronic device on the decoding side. It is a block diagram which shows the example of an encoder and a decoder. 2 is a flowchart illustrating one configuration of a method for high-throughput residual encoding. 6 is a flowchart illustrating one configuration of a method for high-throughput residual coding on the decoding side.

  FIG. 4 is a block diagram illustrating an example of an encoder and a decoder.

  System 400 includes an encoder 411 for generating encoded blocks that are decoded by a decoder 412. The encoder 411 and the decoder 412 may communicate through a network.

  The encoder 411 includes an electronic device 421 configured to encode using a high throughput binarization mode. The electronic device 421 may include a processor and memory in electronic communication with the processor, which stores instructions executable by the processor to perform the operations shown in FIGS.

  The decoder 412 includes an electronic device 422 configured to decode using a high throughput binarization mode. Electronic device 422 may include a processor and memory in electronic communication with the processor, which stores instructions that are executable to perform the operations shown in FIGS.

  FIG. 5 is a flow diagram illustrating one configuration of a method for a high-throughput binarization mode in an electronic device.

  At block 511, the electronic device 421 obtains a block of transformed and quantized coefficients (TQC). In the diamond portion 512, the electronic device 421 determines whether the high-throughput binarization mode condition is satisfied. If the condition is not met at diamond 512, then at block 513, electronic device 421 encodes the block by selectively using the regular coding mode and the bypass coding mode (according to the conventional CABAC selection scheme). .

  If the condition is met at diamond 512, at block 514, electronic device 421 encodes the block using the high-throughput binarization mode and bypass encoding mode. In block 515, the electronic device 421 transmits the generated bitstream over the network and / or stores the generated bitstream in the memory device.

  The HTB mode uses a bypass encoding mode to encode level values. In contrast to the regular coding mode, bypass coding omits the computationally expensive context estimation and probability update steps. This is because the bypass encoding mode assumes a fixed probability distribution for the input bin.

  In addition to using bypass coding mode for encoding, in contrast to conventional CABAC, HTB mode uses a simplified positive / negative coding structure for encoding. For example, conventional CABAC requires four subparts for coding, including Greeter_than_1, Greator_than_2, positive / negative information, and Absolute-3.

  FIG. 6 is a flowchart showing one configuration of encoder processing using the high-throughput binarization mode.

  Blocks 612-615 show the operations performed in block 514 in more detail. In block 612, the electronic device 421 applies an absolute value minus one function for each non-zero value and checks the positive / negative of each non-zero value to determine whether it is positive or negative for any non-zero value from the TQC block. And generate level information. For ease of explanation, the values of the 1D array TQC [4, 0, 3, -3, 2, 1, 0, -1, 0,. . . ]think about. By applying an absolute value minus one function to each non-zero value and checking the positive / negative of each non-zero value, the following six combinations of positive / negative and level information are generated. That is, +3, +2, -2, +1, +0, and -0.

  In block 613, the electronic device 421 uses the mapping table to map the input value to each generated combination of positive / negative and level information. An example of the mapping table is shown in FIG. In addition, FIG. 9 shows an equation for determining input values according to blocks 612 and 613.

  At block 614, the electronic device 421 performs adaptive binarization of input values using a plurality of binarization tables, such as, for example, a VLC table of context adaptive variable length coding (CAVLC). Do. An example of the VLC table of CAVLC is shown in FIG. In addition, FIG. 10 shows an equation for updating the binarization table based on previous input information.

  In one example, block 614 may include first using values from the VLC-Table-0 sequence (FIG. 10) to binarize at least the first input value. When the previous value is greater than a given threshold, such as 3, 5, 13, 27, etc., the VLC table number may be monotonically increased. Thus, adaptive binarization after the first monotonic increase may use values from the VLC-Table-1 sequence, and adaptive binarization after the second monotonic increase may be VLC -The value from the Table-2 column may be used.

  At block 615, the electronic device 421 encodes the value resulting from the adaptive binarization using the CABAC bypass encoding mode.

  "High-throughput binarization mode conditions"

  In one example, if the characteristic corresponding to the block of image data is greater than a preset threshold, the high-throughput binarization mode condition is satisfied and the electronic device 421, for example, has a high-throughput binarization such as an HTB mode flag. The mode indicator may be set to a value of 1 (of course, depending on the design priority, this operation may change the default value of the HTB mode flag or leave the HTB mode flag at the default value) May be included).

  In one example, the electronic device 421 determines whether the bit rate for encoding is greater than a preset threshold. If the bit rate is greater than the preset threshold, the high-throughput binarization mode condition is satisfied. In one example, the preset bit rate threshold corresponds to QP16. However, preset threshold values corresponding to different QP values may be used.

  In one example, the determination of whether the high-throughput binarization mode condition (by the electronic device 421 or the electronic device 422) is satisfied is based on the conversion unit level (eg, the level value generated by the conversion unit) of the corresponding block of image data (But not limited to) is based on whether it is greater than a preset threshold.

  In one example, the high-throughput binarization mode condition may be satisfied when the number of level values of the corresponding block of image data whose size is greater than 0 is greater than a preset threshold, such as 8, for example. In another example, the high-throughput binarization mode condition is satisfied when the number of level values of the corresponding block of image data whose magnitude is greater than the first preset threshold is greater than the second preset threshold. In a further example, the high-throughput binarization mode condition is satisfied when the level value of the corresponding block of image data is greater than a preset threshold.

  20A-E show some examples of configurations that may be used for an encoder or decoder in an example system that operates in accordance with at least some of the principles described in the previous two paragraphs. As shown, FIG. 20A shows operations 1611-1616. As shown, FIG. 20B shows operations 1711-1716. FIG. 20C shows processes 1801-1805 and 1814-1820. FIG. 20D shows processes 1901-1905 and 1914-1920. 20E, the process of FIG. 20C up to the process 1816 is performed in FIG. 20E. In operation 1816, if the counter is greater than the threshold, the configuration continues as shown in FIG. 20E.

  In one example, the determination of whether the high-throughput binarization mode condition (by electronic device 421 or electronic device 422) is met is based on whether the slice level of the corresponding block of image data is greater than a preset threshold. is there.

  "High-throughput binarization mode indicator"

  In one example, the electronic device 421 is configured to set a high-throughput binarization indicator such as an HTB mode flag in a header such as a slice header. The high-throughput binarization indicator may be used to determine whether or not the processing shown in FIG. 5 is performed on the block (s) corresponding to the slice header.

  In one example, by setting the HTB mode flag to “1”, in response to the electronic device 421 observing the HTB mode flag value “1”, the block or blocks corresponding to the slice header On the other hand, the processing shown in the flowchart of FIG. 5 is executed. By setting the HTB mode flag to “0”, the electronic device 421 responds to the observation of the HTB mode flag value “0” in accordance with the block (s) corresponding to the slice header according to the conventional CABAC technique. ) Is encoded.

  In addition, the HTB mode flag value may be observed by the electronic device 422 for decoding. In one example, in response to observing the HTB mode flag value “1”, the electronic device 422 performs the HTB on the block (s) corresponding to the slice header according to the process shown in the flowchart of FIG. The block (s) corresponding to the slice header having mode flag value “1” is decoded. In response to observing the HTB mode flag value “0”, the electronic device 422 decodes the block or blocks corresponding to the slice header having the HTB mode flag value “0” in accordance with conventional CABAC techniques. .

  FIG. 7 is a flowchart showing one configuration of a method for the high-throughput binarization mode in the decoding-side electronic device.

  At block 710, the electronic device 422 obtains a bitstream. In block 711, the electronic device 422 recovers a binary symbol from the obtained bitstream.

  In the diamond portion 712, the electronic device 422 determines whether or not the high-throughput binarization mode condition is satisfied. In one example, the determination may include checking a header, such as a slice header, corresponding to the received bitstream. Checking the header may further include checking the slice header corresponding to the resulting bitstream against the value of the high throughput binarization mode indicator. If the condition is not met at diamond 712, at block 713, electronic device 422 decodes the binary symbol by selectively using the regular decoding mode and the bypass coding mode.

  If the condition is satisfied at diamond 712, at block 714, electronic device 421 decodes the binary symbol using the high-throughput binarization mode and bypass decoding mode. The electronic device 422 stores the obtained block of TQC in the memory device and / or restores the video data at block 715.

  FIG. 8 is a flowchart showing one configuration of the decoder processing using the high-throughput binarization mode.

  Blocks 812-815 show the operations performed in block 714 in more detail. At block 812, the electronic device 422 bypass decodes the encoded binary symbol. In block 813, the electronic device 422 debinates the bypass decoding result. At block 814, the electronic device 422 uses the mapping table to map the input values recovered from the devinarization to the positive and negative and level information. At block 815, the electronic device 422 decodes the transformed and quantized block of coefficients (TQC) using the positive and negative and level information.

  In one example, an electronic device is provided that includes a processor and a memory in electronic communication with the processor. The memory stores instructions that can be executed by the processor to perform operations.

  In one example, one operation may include obtaining a block of transformed and quantized coefficients (TQC). Another operation may include determining whether a high throughput binarization mode condition is met. Another operation may include generating a first bitstream using the high throughput binarization mode in response to determining that the high throughput binarization mode condition is met. Another operation may include generating a second bitstream in response to determining that the high-throughput binarization mode condition is not met. Another operation may include transmitting the generated first or second bitstream to a decoder.

  In one example, the generation of the first bitstream using the high throughput binarization mode may include additional operations. One operation generates positive / negative and level information for any non-zero value from the block by applying an absolute value minus one function for each non-zero value and checking the positive / negative of each non-zero value. Steps may be included. Another operation may include mapping the input value to each generated combination of positive and negative and level information using a mapping table. Another operation may include performing adaptive binarization of the mapped input values using a plurality of binarization tables. Another operation may include encoding the result of adaptive binarization.

  In one example, the plurality of binarization tables includes a CAVLC VLC table. The step of encoding the result of the adaptive binarization may further include an operation using the CABAC bypass encoding mode.

  In one example, adaptive binarization of mapped input values using multiple binarization tables may include additional operations. One operation may include determining whether one of the mapped input values is greater than a preset threshold. Another operation may include performing a table update in response to determining that the mapped input value is greater than a preset threshold. In one example, table update selection includes selection of a table from a set of tables.

  In one example, the generation of the first bitstream may include additional operations. One operation may include encoding the block by selectively using a regular coding mode and a bypass coding mode according to CABAC. Another operation may include generating the first bitstream using only the bypass coding mode.

  In one example, the determination of whether the high-throughput binarization mode condition is satisfied is based on whether the characteristic corresponding to the block of image data is greater than a preset threshold.

  In one example, the determination of whether the high-throughput binarization mode condition is satisfied is based on whether the slice level of the corresponding block of image data is greater than a preset threshold.

  In one example, the determination of whether the high-throughput binarization mode condition is satisfied is based on whether the conversion unit level of the corresponding block of image data is greater than a preset threshold.

  "Lossless coding technology for CABAC in HEVC"

  FIG. 12 is a block diagram illustrating an example of an encoder and a decoder for the lossless coding technique.

  System 1400 includes an encoder 1411 for generating encoded blocks that are decoded by a decoder 1412. The encoder 1411 and the decoder 1412 may communicate through a network.

  The encoder 1411 includes an electronic device 1421 configured to encode using lossless encoding techniques for CABAC in HEVC. Electronic device 1421 may include a processor and memory in electronic communication with the processor, which stores instructions executable by the processor to perform the operations illustrated in FIGS. 13, 16 and 18. .

  The decoder 1412 includes an electronic device 1422 configured to decode using lossless encoding techniques for CABAC in HEVC. Electronic device 1422 may include a processor and memory in electronic communication with the processor, which stores instructions that are executable to perform the operations shown in FIGS.

  FIG. 13 is a flow diagram illustrating one configuration of a method for lossless encoding in an electronic device.

  In block 911, the electronic device 1421 obtains a block representing the residual sample. In one example, the zigzag scan direction may be redefined to match the direction of intra prediction used to remove spatial redundancy between neighboring pixels. There are several intra prediction modes available in the lossless intra coding mode. In one example, in the vertical intra prediction mode, the upper pixel is the predicted value of the current pixel value, and the difference between the current value and the predicted value (upper pixel value in the vertical mode) is the residual sample value. The context model selection may also depend on the direction of intra prediction and the corresponding block size.

  At block 912, the electronic device 1421 generates a significance map for use in the sequence of syntax elements. In block 913, the electronic device 1421 enters a value corresponding to the level of the final position of the block in the significance map field corresponding to the final scan position of the block.

  In block 914, the electronic device 1421 generates a sequence of syntax elements that includes a significance map having the values. The step of generating a sequence of syntax elements excludes the final position encoding step of the conventional CABAC lossless encoding mode.

  FIG. 14 is a table showing columns of syntax elements according to the configuration shown in FIG.

  In contrast to the sequence of syntax elements shown in FIG. 11, some differences may be observed in the sequence of syntax elements shown in FIG. Since the conventional CABAC lossless coding mode includes a final position coding step, the sequence of syntax elements shown in FIG. 11 includes a Last_position_X field and a Last_position_Y field. In contrast, since the configuration of FIG. 14 omits the final position encoding step, the sequence of syntax elements shown in FIG. 14 does not include the Last_position_X field and the Last_position_Y field.

  Both syntax element columns contain significance maps, but there are differences in these significance maps. In the significance map of the column of syntax elements in FIG. 11, the significance map field is not input to correspond to the input field of Last_position_X / Last_position_Y. On the other hand, in FIG. 14, a value corresponding to the level of the final position of the block, that is, “0” in this block example, is input to the significance map field corresponding to the final scan position of the block.

  FIG. 15 is a flowchart showing one configuration of a method for lossless decoding in the electronic device on the decoding side.

  At block 1011, the electronic device 1422 restores from the bitstream a sequence of syntax elements having a significance map field that includes a number of values corresponding to the last scan position of the block. In block 1012, the electronic device 1422 uses the significance map and decodes the level of the block using the value of the significance map. In block 1013, the electronic device 1422 stores the block corresponding to the obtained residual value in the memory device and / or restores the video data.

  FIG. 16 is a flow diagram illustrating another configuration of a method for lossless encoding in an electronic device.

  In block 1111, the electronic device 1421 obtains a string of syntax elements representing level information for the block of residual samples. In block 1112, the electronic device 1421 performs adaptive binarization on the value of the Absolute-3 portion of the syntax element column using a plurality of binarization tables such as, for example, the CAVLC VLC table (FIG. 10). Here, the value of the Absolute-3 portion of the syntax element column is used as an input value for a plurality of binarized tables. The equation for updating the binarization table based on previous input information is shown below.

Listing 1

“I” represents the scan position, and “vlc” represents the current vlc table number.
* Since there is no previous "Absolute-3" available, vlc is initially set to 0 (or 1 for intra slice) and vlc table update is stopped when vlc is equal to 4.

  In one example, block 1111 first uses the values from the VLC-Table-0 sequence for the inter slice (FIG. 10) and the VLC-Table-1 sequence for the intra slice first to binarize at least the first input value. The step of using may be included. When the previous value is greater than a given threshold, such as 3, 5, 13, 27, etc., the VLC table number may be monotonically increased. Thus, adaptive binarization after the first monotonic increase may use values from the VLC-Table-1 sequence, and adaptive binarization after the second monotonic increase may be VLC -The value from the Table-2 column may be used.

  In block 1113, the electronic device 1421 encodes the value resulting from the adaptive binarization using the CABAC bypass encoding mode.

  FIG. 17 is a flowchart showing another configuration of a method for lossless encoding in the decoding-side electronic device.

  In block 1211, the electronic device 1422 recovers a binary symbol from the bitstream. At block 1212, the electronic device 1422 bypass decodes the binary symbol. In block 1213, the electronic device 1422 adaptively debinates the bypass decoding result. In block 1214, the electronic device 1422 reconstructs a block representing residual information using the result of adaptive devinarization.

  FIG. 18 is a flowchart illustrating yet another configuration of a method for lossless encoding in an electronic device.

  In block 1311, the electronic device 1421 accesses only a subset of the CABAC context model. The number of CABAC context models may be 184. To generate the subset, these context models may be filtered based on the relevant characteristics of the context model, for example which context model is related to the frequency component, which context model is related to the scan position, or It may be filtered based on which context model is associated with the CABAC final position encoding step, etc., or any combination thereof. In one example, filtering may be performed by electronic device 1421, while in other examples, a subset is provided to electronic device 1421 so that electronic device 1421 can access the subset provided for lossless coding mode. May be. In one example, to generate the subset, the CABAC context model may be classified based on the relevant characteristics of the context model, eg, which context model is related to the frequency component, which context model is related to the scan position Or which context model is associated with the CABAC final position encoding step, etc., or any combination thereof. In one example, the frequency components and scan positions may be equal and interchangeable.

  In one example, the subset may not include a CABAC context model that has a frequency component that is not equal to the first frequency component. In one example, the resulting subset is 26 context models, ie two context models for encoding significance maps (one for the first luminance frequency component and the other for the first chrominance frequency component). The first luminance frequency component of the Greeter_than_1 flag, the first chrominance frequency component of the Greeter_than_1 flag is encoded, and the first luminance frequency component of the Greeter_than_2 flag of the luminance is encoded. And six context models each for encoding the first chrominance frequency component of the Greeter_than_2 flag. Therefore, a total of 24 context models are used for Greator_than_1 and Greather_than_2. In one example, the first frequency component is accessed only when encoding a significance map at block 1312.

  As shown in Table (3), the six context models for encoding the first luminance frequency of the flag may depend on the sub-block type and the LargeT1 value, where the LargerT1 value is The number of coefficient level values greater than 1 in each sub-block. In one example, the term “sub-block” refers to the subdivision of the residual sample (or block to TQC). For example, for a 4 × 4 sub-block size, a residual sample having a size of 8 × 8 is divided into four 4 × 4 sub-blocks. Similarly, for an 8 × 4 sub-block size, a residual sample having a size of 32 × 8 is divided into eight 8 × 4 sub-blocks. The sub-block is identified by the coding order, where sub-block 0 indicates the first coding sub-block. In one example, the first encoding sub-block is a sub-block located at the lower right of the block. In another example, the first encoding sub-block is a sub-block located at the center of the block.

  At block 1312, the electronic device 1421 encodes the significance map using the subset of context models.

  As described in paragraph 0120, a context model having a frequency component (or scan position) that is not equal to the first frequency component (or scan position) may not be used in the lossless coding mode. This has the benefit of reducing computational complexity and memory for the lossless coding mode. A first subset of the context model may be used for significance map processing. The second subset of the context model may be used for level encoding, such as Greeter_than_1 encoding and / or Greeter_than_2 encoding, for example. The first subset may be different from the second subset.

  In examples applying at least some of the principles described above, the first subset of context models used for significance map processing may include only one context model. In another example applying at least some of the principles described above, the first subset of context models used for significance map processing is based on color information (luminance / color difference), for example two or three context models May include more than one context model. In yet another example applying at least some of the principles described above, the first subset of context models used for significance map processing is based on a prediction type such as, for example, the use of intra-frame or inter-frame prediction within a block. Thus, it may include more than one context model, for example some context models. In another example applying at least some of the principles described above, the first subset of context models used for significance map processing is based on block size, for example two, such as two or three context models The above context model may be included. In another example applying at least some of the principles described above, the first subset of context models used for significance map processing is based on sub-block types, eg 2 or 3 context models, 2 It may contain more than one context model.

  In an example applying at least some of the above principles, the second subset of context models used for level encoding may include only one context model. In another example applying at least some of the principles described above, the second subset of context models used for level encoding is based on color information (luminance / color difference), for example two or three context models, etc. More than one context model may be included. In yet another example applying at least some of the principles described above, the second subset of context models used for level encoding is based on block prediction types, such as the use of intraframe or interframe prediction within a block, for example. Thus, it may include more than one context model, for example some context models. In another example applying at least some of the principles described above, the first subset of context models used for level encoding is based on block size, for example two or more, such as two or three context models The context model may be included. In another example applying at least some of the principles described above, the first subset of context models used in the level encoding process is based on sub-block types, eg 2 or 3 context models, 2 It may contain more than one context model.

  FIG. 19 is a flowchart showing still another configuration of the method for lossless encoding in the decoding-side electronic device.

  At block 1511, the electronic device 1422 accesses only a subset of the CABAC context model. At block 1512, the electronic device 1422 uses the subset of context models to recover binary symbols from the bitstream. In block 1513, the electronic device 1422 restores the video data using the decoding result.

  In the foregoing, configurations that can be realized by the electronic device 1421 are shown in FIGS. 13, 16 and 18. By configuring the encoder by all these configurations, the encoding performance is improved as compared with the known CABAC lossless encoding mode. However, it is also possible and practical to configure an encoder with any combination less than all of these configurations, such as one of these configurations or any two of these configurations, and such configurations are also known. Compared with the CABAC lossless coding mode, the coding performance is improved.

  In the foregoing, configurations that can be implemented by the electronic device 1422 are shown in FIGS. By configuring the decoder by all these configurations, the encoding performance is improved as compared with the known CABAC lossless encoding mode. However, it is also possible and practical to configure the decoder with any combination less than all of these configurations, such as one of these configurations or any two of these configurations, and such configurations are also known. Compared with the CABAC lossless coding mode, the coding performance is improved.

  In one example, an electronic device is provided that includes a processor and a memory in electronic communication with the processor. The memory stores instructions that can be executed by the processor to perform operations.

  In one example, one operation may include obtaining a block representing residual samples for lossless coding. Another operation may include generating a significance map, which includes entering a value corresponding to the level of the block's final scan position in the significance map field corresponding to the block's final scan position. Including the steps of: Another operation may include generating a sequence of syntax elements that includes a significance map having that value. Another operation may include sending a bitstream representing a sequence of generated syntax elements to a decoder.

  In one example, the sequence of syntax elements is generated without performing a final position encoding step of context adaptive binary arithmetic coding (CABAC).

  In one example, another operation may include performing execution-adaptive binarization using a plurality of binarization tables, wherein the value of the Absolute-3 portion of the syntax element column is a plurality of 2 Used as an input value for the value table. Another operation may include encoding the result of adaptive binarization. The plurality of binarization tables may be CAVLC VLC tables.

  In one example, encoding the result of adaptive binarization may include additional operations. Additional operations may include using the CABAC bypass coding mode.

  In one example, adaptive binarization of input values using a plurality of binarization tables may include additional operations. Additional operations may include determining whether one of the input values is greater than a preset threshold. Additional operations may include performing a table update in response to determining that the input value is greater than a preset threshold.

  In one example, another operation may include accessing only a subset of the CABAC context model. Another operation may include encoding a significance map using a subset of the context model. The subset may include a CABAC context model having a frequency component that is not equal to the first frequency.

  In one example, an electronic device is provided that includes a processor and a memory in electronic communication with the processor. The memory stores instructions that can be executed by the processor to perform operations.

  In one example, one operation may include obtaining a block representing residual samples for lossless coding. Another operation may include generating a sequence of syntax elements to represent the block. Another operation may include performing adaptive binarization using a plurality of binarization tables, wherein the value of the Absolute-3 portion of the syntax element column is for a plurality of binarization tables. Used as input value. Another operation may include encoding the result of adaptive binarization. Another operation may include sending this encoding to a decoder.

  In one example, the multiple binarization tables are CAVLC VLC tables.

  In one example, encoding the result of adaptive binarization may include additional operations. Additional operations may include using a context adaptive binary arithmetic coding (CABAC) bypass coding mode.

  In one example, adaptive binarization of input values using a plurality of binarization tables may include additional operations. Additional operations may include determining whether one of the input values is greater than a preset threshold. Additional operations may include performing a table update in response to determining that the input value is greater than a preset threshold.

  In one example, another operation may include generating a significance map, the generating step corresponding to the level of the block's final scan position in the significance map field corresponding to the block's final scan position. Including the step of entering a value. Another operation may include generating a sequence of syntax elements using the generated significance map.

  In one example, the sequence of syntax elements is generated without performing a CABAC final position encoding step.

  In one example, a method is provided. This method may be performed using a decoder. One operation of the method may include filtering a context adaptive binary arithmetic coding (CABAC) context model based on which context model is associated with a frequency component. Another operation of the method may include obtaining a bitstream. Another operation of the method may include recovering a binary symbol from the bitstream. Another operation of the method may include decoding binary symbols using a filtered context model. Another operation of the method may include restoring video data using the decoding result.

  In one example, another operation may include restoring from the bitstream a sequence of syntax elements having a significance map that is populated with a value corresponding to the last scan position of the block representing the residual sample. Another operation may include using a significance map and decoding the level of the block using the value of the significance map.

  In one example, block level decoding may be performed without performing a CABAC final position decoding step.

  In one example, another operation may include bypass decoding the recovered binary symbol. Another operation may include debinarizing the result of bypass decoding adaptively. Another operation may include restoring a block representing residual information using the result of the debinarization.

  In one example, another operation may include using CAVLC multiple VLC tables for adaptive devinarization.

  In one example, bypass decoding may include using a CABAC bypass decoding mode.

  FIG. 21 is a flowchart illustrating one configuration of a method for determining whether a high throughput mode condition is satisfied in a decoding-side electronic device.

  At block 2611, the electronic device 422 obtains a bitstream. At block 2612, the electronic device 422 obtains a block of level values. In one example, the block includes a TQC block.

  At block 2613, the electronic device 422 determines the number of level values that are not equal to zero. In diamond portion 2614, electronic device 422 determines whether the number is greater than a preset threshold. In one example, the preset threshold may be 8, which is half the number of values in a 4 × 4 block. In an example with a block size having an N level value, the threshold may correspond to 50% of N. In one example, the electronic device 422 receives signaling from the electronic device 421. The signaling sent by the electronic device 421 may include information that may be used by the electronic device 421 to specify a preset threshold or to define a preset threshold.

  At diamond 2614, if the number is less than or equal to the preset threshold, at block 2615, electronic device 422 decodes a level value not equal to 0 by a first binarization method. In diamond 2614, if the number is greater than the preset threshold, then in block 2616, electronic device 422 determines a level value not equal to 0 by a second binarization method different from the first binarization method. Decrypt. In one example, the second binarization method may include a high throughput devinarization mode, such as the HTB mode described above. In one example, the first binarization method may include conventional CABAC binarization.

  FIG. 22 is a flowchart illustrating another configuration of a method for determining whether a high-throughput mode condition is satisfied in a decoding-side electronic device.

  At block 2711, the electronic device 422 obtains a bitstream. At block 2712, the electronic device 422 obtains a block of level values. In one example, the block includes a TQC block.

  At block 2713, the electronic device 422 determines the number of level values that have an absolute value greater than the first preset threshold. In one example, the first preset threshold may be 1 or 2, but in other examples other first preset thresholds may be used. In diamond portion 2714, electronic device 422 determines whether the number is greater than a second preset threshold. In one example, the second preset threshold may be 8, which is half the number of values in a 4 × 4 block. In an example with a block size having an N level value, the second preset threshold may correspond to 50% of N.

  In one example, the electronic device 422 receives signaling from the electronic device 421. The signaling sent by the electronic device 421 may specify a first preset threshold and / or a second preset threshold, or may be sent by the electronic device 421 to define the first preset threshold and / or the second preset threshold. It may contain information that can be used.

  In diamond 2714, if the number is less than or equal to the second preset threshold, in block 2715, electronic device 422 decodes the level value not equal to 0 using the first binarization method. In diamond 2714, if the number is greater than the preset threshold, then in block 2716, electronic device 422 determines a level value not equal to 0 by a second binarization method different from the first binarization method. Decrypt. In one example, the second binarization method may include a high throughput devinarization mode, such as the HTB mode described above. In one example, the first binarization method may include binarization of known CABAC decoding.

  FIG. 23 is a block diagram illustrating an example of an encoder and a decoder.

  System 2400 includes an encoder 2411 for generating encoded blocks that are decoded by a decoder 2412. The encoder 2411 and the decoder 2412 may communicate through a network.

  The decoder 2412 includes an electronic device 2422 configured to decode using high throughput significance map processing. The electronic device 2422 may include a processor and memory in electronic communication with the processor, which stores instructions that are executable to perform the operations shown in FIGS.

  The encoder 2411 includes an electronic device 2421, which may include a processor and a memory in electronic communication with the processor, the memory corresponding to the description of the configurations shown in FIGS. The instructions that are executable by the processor to perform operations as understood by those of ordinary skill in the art from the description to be stored.

  FIG. 24 is a flowchart illustrating one configuration of a method for high-throughput significance map decoding in the decoding-side electronic device.

  At block 2801, the electronic device 2422 obtains a bitstream. In block 2802, the electronic device 2422 obtains a block of level values. In one example, the block includes a TQC block. In block 2803, the electronic device 2422 obtains a block level value, such as a block first level value or a block next level value.

  In diamond 2804, electronic device 2422 determines whether the obtained level value is the final level value of the block. In diamond 2804, if the obtained level value is not the final level value, electronic device 2422 proceeds to diamond 2814. In diamond 2804, if the obtained level value is the final level value, in block 2805, the electronic device 2422 decodes the magnitude of the level value (this operation includes positive / negative information and absolute value for each level value). Including steps to define both).

  Referring to the diamond portion 2814 again, the electronic device 2422 uses the first decoding method to determine whether the obtained level value is not zero. In diamond 2814, if the obtained level value is not 0, electronic device 2422 proceeds to block 2815. Otherwise, electronic device 2422 returns to block 2803. At block 2815, the electronic device 2422 increments the counter.

  In diamond portion 2816, electronic device 2422 determines whether the current count of the counter is greater than a preset threshold. In one example, the preset threshold value may include the preset threshold value described with reference to FIG. At diamond 2816, if the current count of the counter is greater than the preset threshold, electronic device 2422 proceeds to block 2817. Otherwise, electronic device 2422 returns to block 2803.

  In block 2817, the electronic device 2422 obtains the next level value of the block. In diamond 2818, electronic device 2422 determines whether the obtained level value is the final level value of the block. In diamond 2818, if the obtained level value is not the final level value, electronic device 2422 proceeds to block 2819. Otherwise, at block 2820, the electronic device 2422 decodes the magnitude of the level value.

  At block 2819, the electronic device 2422 uses a second decoding method that is different from the first decoding method to determine whether the resulting level value is non-zero. In one example, the second decoding method includes a high-throughput decoding method or a bypass decoding method. In one example, the first decoding method includes a CABAC regular decoding mode.

  In accordance with the above, the significance map may be decoded element by element, eg, by significance map field. When the preset threshold is reached, the electronic device 2422 may change the decoding of the remaining significance map portions. A high throughput or bypass significance map decoding mode may be used for the remaining significance map portions. Therefore, decoding performance can be improved over conventional CABAC significance map decoding.

  FIG. 25 is a flowchart illustrating another configuration of a method for high-throughput significance map decoding in the decoding-side electronic device.

  In the method shown in FIG. 25, processes 2901-2905 similar to processes 2801-2805 (FIG. 24) may be performed as shown. In the diamond portion 2914, the electronic device 2422 uses the first decoding method to determine whether the absolute value of the obtained level value is greater than the first threshold value. In one example, the first threshold may be 1 or 2, but in other examples other first thresholds may be used. In diamond 2914, if the absolute value of the obtained level value is greater than the first threshold, electronic device 2422 proceeds to block 2915. Otherwise, electronic device 2422 returns to block 2903. At block 2915, the electronic device 2422 increments the counter.

  In diamond portion 2916, electronic device 2422 determines whether the current count of the counter is greater than a second preset threshold. In one example, the second preset threshold may be 8, which is half the number of values in a 4 × 4 block. In an example with a block size having an N level value, the second preset threshold may correspond to 50% of N. In diamond 2916, if the current count of the counter is greater than the second threshold, electronic device 2422 proceeds to block 2917. Otherwise, electronic device 2422 returns to block 2903.

  In block 2917, the electronic device 2422 obtains the next level value of the block. In diamond portion 2918, electronic device 2422 determines whether the obtained level value is the final level value of the block. In diamond 2918, if the obtained level value is not the final level value, electronic device 2422 proceeds to block 2919. Otherwise, at block 2920, the electronic device 2422 decodes the magnitude of the level value.

  In block 2919, the electronic device 2422 uses a second decoding method different from the first decoding method to determine whether the absolute value of the obtained level value is greater than the first threshold value. In one example, the second decoding method includes a high-throughput decoding method or a bypass decoding method. In one example, the first decoding method includes a CABAC regular decoding mode.

  In accordance with the above, the significance map may be decoded element by element, eg, by significance map field. When the preset threshold is reached, the electronic device 2422 may change the decoding of the remaining significance map portions. A high throughput or bypass significance map decoding mode may be used for the remaining significance map portions. Therefore, decoding performance can be improved over conventional CABAC significance map decoding.

  FIG. 26 is a flowchart illustrating one configuration of a method for high-throughput significance map decoding with a decoding bypass function in the decoding-side electronic device.

  In the method shown in FIG. 26, processes 3001 to 3004 and 3014 to 3016 similar to processes 2801 to 2804 and 2814 to 2816 (FIG. 24) may be performed as illustrated. At block 3005, the electronic device 2422 restores the magnitude of the level value using a third decoding method such as a binarization method. At block 3020, the electronic device 2422 restores the magnitude of the first portion of the level value using a third decoding method, and uses a fourth decoding method such as a different binarization method to obtain the level value. The size of the second part of is restored.

  In one example, the first part of the level value includes the level value processed by the first decoding method. The second part of the level value includes the level value that has not been processed by the first decoding method.

  It should be apparent that other configurations of the method for high throughput significance map decoding with decoding bypass functionality in the decoding-side electronic device similar to the configuration shown in FIG. 26 are possible and practical. For example, in another configuration, the electronic device 2422 uses the first decoding method to determine whether the absolute value of the obtained level value is greater than the first preset threshold, similar to the diamond portion 2914 (FIG. 25). judge. In addition, the electronic device 2422 determines whether the counter is greater than the second preset threshold, similar to the diamond portion 2916 (FIG. 25).

  In accordance with the above, the significance map may be decoded element by element, eg, by significance map field. When the preset threshold is reached, the electronic device 2422 may stop decoding the significance map (the remaining elements of the significance map are not decoded). The level values corresponding to the decoded elements are then processed using a binarization method (eg, a binarization method that can send a value of 0), while the remaining elements are processed by different binarization methods ( For example, a binarization method that cannot send a value of 0). Therefore, decoding performance can be improved over conventional CABAC significance map decoding.

  FIG. 27 is a flowchart showing one configuration of a method for high-throughput significance map decoding with a decoding method switching function in the decoding-side electronic device.

  In the method shown in FIG. 27, processes 3801 to 3804 and 3814 to 3819 similar to processes 2801 to 2804 and 2814 to 2819 (FIG. 24) may be performed as shown. In block 3805, the electronic device 2422 restores the magnitude of the level value using the third decoding method (third decoding method of FIG. 26). At block 3820, the electronic device 2422 restores the magnitude of the first portion of the level value using the third decoding method and uses the fourth decoding method (fourth decoding method of FIG. 26). Restore the size of the second part of the level value. In one example, the first portion of the level value includes the level value obtained at block 3803, while the second portion of the level value includes the level value obtained at block 3817.

  It should be apparent that other configurations of the method for high throughput significance map decoding with a decoding bypass function in the decoding-side electronic device similar to the configuration shown in FIG. 27 are possible and practical. For example, in another configuration, the electronic device 2422 uses the first decoding method to determine whether the absolute value of the obtained level value is greater than the first preset threshold, similar to the diamond portion 2914 (FIG. 25). judge. In addition, the electronic device 2422 determines whether the counter is greater than the second preset threshold, similar to the diamond portion 2916 (FIG. 25).

  In one example, a first electronic device is provided that includes a processor and a memory in electronic communication with the processor. The memory stores instructions that can be executed by the processor to perform operations.

  In one example, one operation may include receiving a bitstream. Another operation may include obtaining a block of level values based on the received bitstream. Another operation may include identifying a portion of the level value according to a threshold value. Another operation may include processing any remaining values of the level value using the high throughput significance map processing mode after identifying the portion. Another operation may include restoring video data based on the processing.

  In one example, a second electronic device is provided that includes a processor and a memory in electronic communication with the processor. The memory stores instructions that can be executed by the processor to perform operations. One operation may include sending signaling to the first electronic device, where the signaling specifies a threshold value.

  In accordance with the above, the significance map may be decoded element by element, eg, by significance map field. When the preset threshold is reached, the electronic device 2422 may stop decoding the significance map (the remaining elements of the significance map are not decoded). The level values corresponding to the decoded elements are then processed using a binarization method (eg, a binarization method that can send a value of 0), while the remaining elements are processed by different binarization methods ( For example, a binarization method that cannot send a value of 0). Therefore, decoding performance can be improved over conventional CABAC significance map decoding.

  “Lossless coding with different parameter selection techniques for CABAC in HEVC”

  When using CABAC coding in HEVC in lossless coding mode, encoding / decoding is computationally complex. One reason for this computational complexity is the encoding of the syntax element “Absolute-3”. In known CABAC encoding, an exponential Golomb-Rice encoding method is used to encode syntax elements.

  As a background, the exponential Golomb-Rice (GR) coding method uses the Rice parameter update table shown in FIG. As described in more detail in the next paragraph, the GR encoding method is used to encode the syntax element “Absolute-3” (ie, the last row of the table of FIG. 2) in CABAC's known lossless encoding mode. Applies.

  Rice parameters control the conversion from symbol to bin. For purposes of illustration, symbols 0, 11, 4,..., Using the table of FIG. . . Think about converting. Here, “0” (first symbol) is the first symbol of the sub-block. Since the first symbol is the first symbol of the sub-block, the Rice parameter is initialized to 0 for the first symbol. The first symbol “0” is encoded using 0 of the current Rice parameter. In one example, the process of encoding a symbol with the RP's Rice parameter includes the steps of calculating the value of Quuient = floor ((symbol-1) / RP), and a bin string equal to 0 after a Quotient sequence of bins equal to 1. Generating an output including. Here, quotient is an integer, and floor () is an operation for mapping a value including an integer and a fractional component to the integer component. For illustration, encoding the symbol “5” with Rice parameter 3 results in a quotient value of 1 and an output bin of “01”. Similarly, encoding the symbol “100” with the Rice parameter 33 results in a quotient value of 3 and an output bin “0001”. In an alternative example, the process of encoding a symbol with the RP's Rice parameter comprises the steps of calculating the value of Quuent = floor ((symbol-1) / RP) and a bin equal to 1 after a quaint column of bins equal to 0. Generating an output including. In yet another example, the process of encoding a symbol with the RP's Rice parameter comprises selecting an RP-th lookup table that defines a mapping between symbols and bin columns from a set of lookup tables. When the result of the lookup according to the table of FIG. 28 is 0, the Rice parameter does not update the next symbol. Therefore, the second symbol “11” is encoded using the current Rice parameter 0. When the result of the lookup for the second symbol “11” and the rice parameter “0” (“2”) is different from the current rice parameter value (ie, 0), the rice parameter is updated from 0 to 2. The third symbol “4” is then encoded with the current Rice parameter 2. When the result of the lookup is not a different value from the current Rice parameter, Rice parameter 2 is used for the next symbol.

  Encoding / decoding may consume a significant amount of processing resources and / or be completed, as the GR encoding of “Absolute-3” values according to known CABAC is computationally complex It can take a considerable amount of time. The present disclosure below addresses this and other problems.

  FIG. 29 is a block diagram illustrating an example of an encoder and a decoder.

  System 2900 includes an encoder 2911 for generating encoded blocks that are decoded by a decoder 2912. The encoder 2911 and the decoder 2912 may communicate through a network.

  The encoder 2911 includes an electronic device 2921 configured to encode using lossless coding with different parameter selections for CABAC in HEVC. Electronic device 2921 may include a processor and memory in electronic communication with the processor, which stores instructions executable by the processor to perform the operations shown in FIG.

  The decoder 2912 includes an electronic device 2922 configured to decode using lossless coding with different parameter selections for CABAC in HEVC. Electronic device 2922 may include a processor and memory in electronic communication with the processor, which stores instructions executable to perform the operations shown in FIG.

  FIG. 30 is a flow diagram illustrating one configuration of a method for lossless encoding with different parameter selections in an electronic device.

  At block 3011, the electronic device 2921 obtains a block of data that is encoded using an arithmetic-based encoder, such as an encoder based on CABAC. In diamond 3012, electronic device 2921 determines whether the block is encoded using lossless encoding. If the block is not encoded using lossless encoding, at block 3013, the electronic device 2921 encodes the block of data using the first Absolute-3 encoding technique.

  If the block is to be encoded using lossless encoding, at block 3014, the electronic device 2921 encodes the block of data using a second, different Absolute-3 encoding technique. At block 3015, the electronic device 2921 transmits the generated bitstream over the network and / or stores the generated bitstream in the memory device.

  In one example, the first Absolute-3 encoding technique includes the CABAC encoded RG encoding technique. That is, the Rice parameter is initialized to 0 at each sub-block encoding stage, and the five Rice parameters in the table shown in FIG. 28 are considered. In one example, the second different Absolute-3 encoding technique does not initialize to 0 in each sub-block encoding stage, ie is initialized differently and / or, for example, reduced Rice parameter update Use different rice parameter update tables, such as tables.

  In one example, the different initializations may include initializing Rice parameters to 0 in each block rather than in each sub-block. In one example, the different initializations may include using the final rice parameters used in the previous sub-block as the initial rice parameters for the current sub-block.

  In one example, the different initializations may include initializing based on residual sample statistics. In one example, the different initializations may include initializing with predefined Rice parameter values based on block type, block size, color information (luminance / color difference), etc., or any combination thereof. Good. The block type is a value for representing a block on the basis of the block size of the block, block prediction information (intra / inter), and block color information (luminance / color difference). In one example, the different initialization is a predefined value “1” when the current block type is equal to a particular predefined value or values such as “2” and / or “5”, for example. A step of initializing the rice parameters.

  In one example, the different Rice parameter update table includes fewer Rice parameters than the Rice parameter update table used for the first Absolute-3 encoding technique. In one example, the different Rice parameter update tables include only the first two cases (Rice parameters equal to “0” and “1”). An example of such a rice parameter update table is included in FIG.

  In one example, if a second different Absolute-3 encoding technique is used, the electronic device 2921 sets the corresponding indicator to a value of 1, for example, a flag associated with the second different Absolute-3 encoding technique. (This operation may of course include changing the default value of the flag or leaving the flag at the default value, depending on design priority).

  FIG. 31 is a flow diagram illustrating one configuration of a method for lossless encoding with different parameter selection in the decoding-side electronic device.

  At block 3110, the electronic device 2922 obtains a bitstream. In block 3111, the electronic device 2922 restores a binary symbol from the obtained bitstream.

  At diamond 3112, electronic device 2922 determines whether the binary symbol is decoded using lossless decoding. In one example, this determination may include checking a header, such as a slice header, corresponding to the received bitstream. Checking the header may further include checking the slice header corresponding to the resulting bitstream for the value of a flag associated with a second different Absolute-3 encoding technique. In another example, this determination may include checking a previously decoded symbol associated with the block, such as a block type or a quantization parameter that controls the conversion from coefficient level to TQC. If the condition is not met at diamond 3112, at block 3113, electronic device 2922 obtains a block of TQC using the first Absolute-3 encoding technique.

  If the condition is satisfied at diamond 3112, at block 3114, electronic device 2921 obtains residual samples using a second, different Absolute-3 encoding technique. In block 3115, the electronic device 2922 stores the obtained block of TQC or the obtained residual sample in a memory device and / or recovers the video data.

  "High Throughput Coding for CABAC in HEVC"

  When using CABAC encoding in HEVC, it depends on different factors such as, but not limited to, the total number of bins / pixels, the number of bypass bins / pixels, and the number of regular (or context) encoding bins / pixels. Throughput performance can vary. Thus, depending on these factors, encoding can consume a significant amount of processing resources and / or can take a significant amount of time. The present disclosure below addresses this and other problems.

  As background, according to the known CABAC, up to 25 level code flags of syntax elements are context encoded. The remaining level code flags are bypass encoded. A predefined (and fixed) number of Greeter_than_1 flags, i.e. 8 Greeter_than_1 flags, are context encoded. A predefined (and fixed) number, ie one Greeter_than_2 flag, is context encoded. All of the significance map flags, i.e. up to 16, are context encoded (depending on the final position information of the block, the syntax element may have a significance map flag of less than 16. Thus, a given A maximum of 25 context coding bins are required for the subset block (25 bins / 16 pixels = 1.56 bins / pixel) The above example is for using 4 × 4 sub-blocks.

  FIG. 34 is a flowchart illustrating one configuration of a method for high-throughput encoding for CABAC in HEVC in an electronic device.

  At block 3411, the electronic device 3321 obtains a block of data to be encoded using an arithmetic-based encoder, such as an encoder based on CABAC. At block 3412, the electronic device 3321 context encodes a first amount of level code flags in the syntax element, such as, for example, the CABAC syntax element Greator_tan_1 and Greator_than_2 flags. The first quantity includes a first predefined number, such as 9, for example, 8 Greater_than_1 flag and 1 Greater_than_2 flag.

  At block 3413, the electronic device 3321 identifies the number of actually encoded bins in the significance map of the syntax element. At block 3414, the electronic device 3321 determines the difference between the second predefined number, eg, 16 in CABAC, and the specified number. At block 3415, the electronic device 3321 context encodes a second quantity of level code flags, where the second quantity includes a determined difference. At block 3416, the electronic device 3321 transmits the generated bitstream over the network and / or stores the generated bitstream in a memory device.

  In one example, if the configuration shown in FIG. 34 is used, the electronic device 3321 may set a corresponding indicator, such as a flag, to a value of 1 (this operation, of course, depending on the design priority, May include changing the default value of, or leaving the flag at the default value). In one example, the indicator may specify the number of additional Greater_than_1 and / or Greater_than_2 flags that are context encoded.

  An example of syntax elements generated according to the above configuration is shown in FIG. In this example, the number of bins actually encoded in the significance map of the example syntax element is twelve. The defined difference between 16 and 12 is 4. The first amount of level code flags that are context encoded is 9 (8 Greater_than_1 flags and 1 Greater_than_2 flag). The second amount of level code flags that are context encoded is four. In this particular example, these four are all Greater_than_1 flags, but in the other examples these four have one or more Greater_than_1 flags and one or more Greater_than_2 flags, or four Greater_than_2 flags. May be included. The remaining level code flags are bypass encoded.

  FIG. 35 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in the electronic device on the decoding side.

  At block 3510, the electronic device 3322 obtains a bitstream. In block 3511, the electronic device 3322 restores a binary symbol from the obtained bitstream.

  At block 3512, the electronic device 3322 context decodes a first quantity level code flag of the syntax element, eg, the CABAC syntax element Greator_tan_1 and Greator_tan_2 flags, where the first quantity is a first pre-defined The number is, for example, nine, ie, 8 Greater_than_1 flag, 1 Greeter_than_2 flag, etc. In diamond portion 3513, electronic device 3322 determines whether the additional level code flag of the syntax element is context encoded. In one example, this determination may include checking a header, such as a slice header, corresponding to the received bitstream. Checking the header may further include checking the slice header corresponding to the resulting bitstream for the value of an indicator such as a flag. If electronic device 3322 determines in diamond 3513 that the additional level code flag is not context encoded, then in block 3514, electronic device 3322 bypass decodes the remaining level code flags of the syntax element.

  If electronic device 3322 determines in diamond 3513 that the additional level code flag is context encoded, then in block 3515, electronic device 3322 uses the second amount of syntax element level code flag in the context. Decrypt. In one example, the electronic device 3322 may determine the number of additional Greater_than_1 and / or Greater_than_2 flags that are context encoded based on information from the slice header. At block 3514, the electronic device 3322 bypass decodes any remaining level code flags. At block 3516, the electronic device 3322 stores the obtained block of TQC or the obtained residual sample in a memory device and / or restores the video data.

  FIG. 38 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in the decoding-side electronic device.

  At block 3850, the electronic device 3322 obtains a bitstream. At block 3851, the electronic device 3322 obtains a block of level values from the resulting bitstream.

  In block 3853, the electronic device 3322 determines the location of the final significant coefficient of the block. At block 3854, the electronic device 3322 context decodes significant coefficients from the resulting block to determine a maximum number of symbols for context decoding as a function of the determined position.

  At block 3855, the electronic device 3322 resets the counter, eg, by setting the counter to zero. In diamond 3856, electronic device 3322 determines whether there are any remaining level code flags to be decoded. If no level code flag remains in diamond 3856, at block 3857 the electronic device 3322 stores the resulting block of TQC or the resulting residual sample in a memory device and / or stores the video data. Restore.

  If the level code flag remains (diamond portion 3856), the electronic device 3322 determines in the diamond portion 3858 whether the counter is greater than the threshold value. In one example, the threshold may be associated with a defined maximum value. In one example, the threshold value may correspond to a difference between a defined maximum value and the number of significance map flags for the block. At diamond 3858, if the counter is greater than the threshold, at block 3859, electronic device 3322 bypass decodes the level code flag. At diamond 3858, if the counter is less than or equal to the threshold, at block 3860, electronic device 3322 context decodes the level code flag. At block 3861, the electronic device 3322 increments the counter.

  FIG. 39 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in the decoding-side electronic device.

  At block 3950, the electronic device 3322 obtains a bitstream. At block 3951, the electronic device 3322 obtains a block of level values from the resulting bitstream.

  In block 3953, the electronic device 3322 determines the location of the final significant coefficient. At block 3954, the electronic device 3322 context decodes significant coefficients from the resulting block to determine a maximum number of symbols for context decoding as a function of both the determined position and sub-block position. Processes 3955 to 3961 can correspond to processes 3855 to 3861.

  FIG. 40 is a flowchart showing one configuration of a method for high-throughput encoding for CABAC in HEVC in the decoding-side electronic device.

  At block 4050, the electronic device 3322 obtains a bitstream. In block 4051, the electronic device 3322 obtains a block of level values from the resulting bitstream.

  In block 4053, the electronic device 3322 determines the location of the final significant coefficient. At block 4054, the electronic device 3322 context decodes significant coefficients from the resulting block and determines a maximum number of symbols for context decoding as a function of block characteristics, eg, the number of significant coefficients of the block. Processing 4055 to 4061 may correspond to processing 3855 to 3861.

  In one example, a system is provided. The system obtains a block of level values from the bitstream, context decodes the block's level code flag, checks if the block's next level code flag exists, and if the next level If the code flag is present, in response to determining whether the count of the context encoded level code flag is greater than the threshold and determining that the count is less than or equal to the threshold, the next level code flag In response to determining that the count is greater than the threshold, context decoding the next level code flag, and using the decoded level code flag, a block or residual sample of TQC And restore the restored block to the memory device Or, and / or configured to perform the method comprising: restoring the video data may include an electronic device.

  The electronic device may be configured to increment the count in response to context encoding the next level code flag. The electronic device may be configured to repeat checking, determining, decoding, and incrementing until all level code flags of the block are decoded. The electronic device may be configured to increment the count in response to context encoding the first level code flag.

  The electronic device may be configured to determine a location of the final significant coefficient of the block and to determine a maximum number of symbols for context decoding based at least in part on the characteristics of the block. The electronic device may be configured to set the threshold according to the result of this maximum number determination.

  The electronic device may be configured to determine a position of the final significant coefficient of the block and to determine a maximum number of symbols for context decoding based at least in part on the determined position. The electronic device may be configured to determine a maximum number of symbols for context decoding based at least in part on the defined location and at least in part on the sub-block location.

  The electronic device context decodes a first quantity level code flag of a syntax element associated with the block, the first quantity being equal to a first predefined number; Identifying the number of actually encoded bins in the significance map of the syntax element, defining a difference between the second predefined number and the identified number, and a syntax element The second amount of level code flags may be context decoded, the second amount including a defined difference and context decoded. In one example, the first predefined number may include nine. In one example, the second predefined number may include 16. In one example, the level code flag corresponding to the first amount of context-encoded level code flag includes eight “greater_than_1” flags and one “greater_than_2” flag. In one example, the level code flag corresponding to the second amount of context-encoded level code flag includes only the “greater_than_1” flag. In one example, the level code flag corresponding to the second amount of context-encoded level code flag includes only the “greater_than_2” flag. In one example, the level code flag corresponding to the second amount of context-encoded flag includes a third predefined number of “glater_than_2” flags and a dynamic number of “glater_than_1” flags; Here, the dynamic number includes the difference between the second quantity and a third predefined number.

  In one example, a system is provided. The system may include a first electronic device of an encoder that obtains a block of data to be encoded using an arithmetic-based encoder and that the block of data performs lossless encoding. In response to determining whether the block of data is not encoded using lossless encoding, and using the first Absolute-3 encoding technique in response to determining whether the block of data is not encoded using lossless encoding. In response to encoding the block and determining that the block of data is encoded using lossless encoding, the block of data is encoded using the second Absolute-3 encoding technique. That is, the second Absolute-3 encoding technique is different from the first Absolute-3 encoding technique. Configured to perform a that stores this coding in a memory device.

  The system may further include a second electronic device of the decoder, the second electronic device determining whether the received binary symbol is decoded using lossless decoding and binary. In response to determining that the symbol is not decoded using lossless decoding, obtaining a block of TQC using the first Absolute-3 coding technique and decoding the binary symbol using lossless decoding And obtaining a residual sample using a second Absolute-3 encoding technique.

  In response to determining that the block of data is not encoded using lossless encoding, the first electronic device initializes the Rice parameter to 0 for the initial value of the sub-block; In response to determining that the block is encoded using lossless encoding, the initial value of the subblock is configured to use the Rice parameter from the final value of the previous subblock. May be.

  In response to determining that the block of data is not encoded using lossless encoding, the first electronic device initializes the Rice parameter to 0 for the initial value of the sub-block; In response to determining that the block is encoded using lossless encoding, the initial value of the sub-block is configured to bypass initializing the Rice parameter to 0 Also good.

  In response to determining that the block of data is encoded using lossless encoding, the first electronic device is out of the group including block type, block size, and color information (luminance / color difference). It may be configured to initialize the rice parameters with predefined values based on at least one selected.

  In response to determining that the block of data is encoded using lossless encoding, the first electronic device initializes the Rice parameter to 1 when the current block type is equal to 2 or 5 It may be configured as follows.

  In response to determining that the block of data is not encoded using lossless encoding, the first electronic device initializes the Rice parameter to 0 for the initial value of the sub-block; In response to determining that the block is encoded using lossless encoding, the initial value of the sub-block is configured to bypass initializing the Rice parameter to 0 Also good.

  In response to determining that the binary symbol is not decoded using lossless decoding, the first electronic device uses the first Rice parameter update table and the binary symbol uses lossless decoding. In response to determining that it is decrypted, a second Rice parameter update table may be used that is different from the first Rice parameter update table.

  The second rice parameter update table may include a shortened version of the first rice parameter update table. In one example, only the second rice parameter update table is configured to prevent an update after the current rice parameter is updated or initialized to 2, 3 or 4.

  In one example, a system is provided. The system may include a first electronic device of an encoder that obtains a block of data to be encoded using an arithmetic-based encoder and a first quantity of syntax elements. Context encoding of the level code flag, wherein the first quantity is equal to the first pre-defined number, and in the context encoding and syntax element significance maps, Identifying the number of encoded bins, determining a difference between a second predefined number and the identified number, and context encoding a second amount of level code flags in the syntax element Where the second quantity includes a defined difference, the context encoding, and the bitstream generated by the context encoding. Configured to perform the be stored in memory devices.

  The arithmetic based encoder may include a CABAC encoder. The first predefined number may include nine. The second predefined number may include 16. The level code flag corresponding to the first amount of context encoded level code flags may include eight “create_than_1” flags and one “create_than_2” flag. The level code flag corresponding to the second amount of context-encoded level code flag may include only the “create_than_1” flag. The level code flag corresponding to the second amount of context-encoded level code flag may include only the “create_than — 2” flag. The level code flags corresponding to the second amount of context-encoded flags may include a third predefined number of “grate_than_2” flags and a dynamic number of “grate_than_1” flags, where The dynamic number includes the difference between the second quantity and the third predefined number.

  “High Throughput Residual Coding for Transform Skip Blocks for CABAC in HEVC”

  FIG. 41 is a block diagram illustrating an example of an encoder and a decoder.

  System 4100 includes an encoder 4111 for generating encoded blocks that are decoded by a decoder 4112. The encoder 4111 and the decoder 4112 may communicate through a network.

  The encoder 4111 includes an electronic device 4121 configured to encode using a high throughput residual encoding mode. Electronic device 4121 may include a processor and memory in electronic communication with the processor, which stores instructions executable by the processor to perform the operations shown in FIG.

  The decoder 4112 includes an electronic device 4122 configured to decode using a high throughput residual encoding mode. Electronic device 4122 may include a processor and memory in electronic communication with the processor, which stores instructions that are executable to perform the operations shown in FIG.

  FIG. 42 is a flow diagram illustrating one configuration of a method for high throughput residual coding.

  At block 4211, the electronic device 4121 obtains a block of data that is encoded using an arithmetic-based encoder, such as a CABAC-based encoder. At diamond 4212, the electronic device 4121 determines whether the block is encoded using high throughput residual encoding. If the block is not encoded using high-throughput residual encoding, at block 4213 the electronic device 4121 encodes the block of data using the first encoding technique. In one example, the first encoding technique may include an Absolute-3 encoding technique, such as the encoding technique described with reference to block 3013 of FIG.

  If the block is to be encoded using high-throughput residual encoding, in block 4214 the electronic device 4121 may block the data using a second encoding technique that is different from the first encoding technique. Is encoded. At block 4215, the electronic device 4121 transmits the generated bitstream over the network and / or stores the generated bitstream in the memory device.

  In one example, the second encoding technique includes only a subset of the encoding stages of the first encoding technique. In one example, the first encoding technique includes a Greater_than_1 encoding stage and a Greator_than_2 encoding stage, and the second encoding technique does not include at least one of the Greeter_than_1 encoding stage and the Greater_than_2 encoding stage.

  In one example, the first encoding technique includes Absolute-3 encoding after significance map encoding, while the second encoding technique does not include Absolute-3 encoding after significance map encoding. In an example of the second encoding technique, the value of Absolute-1 or Absolute-2 is encoded after significance map encoding. In one example, if both the Greeter_than_1 and Greeter_tan_2 encoding steps are skipped, the Absolute-1 value is encoded, and if the Greeter_than_2 encoding step is skipped and the Greeter_than_1 encoding step is not skipped, Absolute- Binary is encoded. In one example, positive / negative encoding may be performed after Absolute-1 or Absolute-2 encoding, while in other examples, positive / negative encoding is performed before Absolute-1 or Absolute-2 encoding. Also good.

  In one example, Absolute-1 or Absolute-2 value encoding uses any Golomb-Rice (GR) code described herein. For example, absolute-1 or absolute-2 value encoding may use the GR encoding technique described with reference to FIG. In one example, Absolute-1 or Absolute-2 binary encoding uses the same GR encoding used for the first encoding technique.

  In one example, the G-R encoding technique used for Absolute-1 or Absolute-2 binary encoding may be block type, block size, or color information (luminance / color difference), etc., or any combination thereof. The step of initializing with pre-defined Rice parameter values may be included. In one example, the predefined rice parameters may depend on the texture (brightness / color difference). For example, the predefined Rice parameter value may be 2 for the luminance block or 1 for the color difference block. In one example, the same G-R encoding is used for Absolute-1 or Absolute-2 encoding and Absolute-3 encoding.

  In one example, the electronic device 4121 may explicitly signal initialization information to the decoding side. The initialization information may include a rice parameter initialization value for the bitstream or the current block, i.e. per block. If the initialization information includes more than one Rice parameter initialization value for a block, the initialization information may further indicate criteria for using the first value or the second value, i.e., The value of may be 2 and 1, and the condition may be 2 for the luminance block and / or 1 for the chrominance block. In one example, the additional information may include additional syntax elements.

  In one example, the diamond portion 4212 may include a step of determining whether a Transform_skip_flag or a Trans_quant_Bypass_flag of a block of data is set. In one example, if Transform_skip_flag or Trans_quant_Bypass_flag is equal to 1, then high-throughput residual encoding is used for that block of data (conversely, if Transform_skip_flag and Trans_quant_block data are both set to Transform_skip_flag and Trans_quant_data For high throughput residual coding).

  As a background, Transform_skip_flag indicates whether or not the corresponding block has been converted. In the known scheme, Transform_skip_flag is equal to 0 when the corresponding block is transformed. When Transform_skip_flag is equal to 1, the corresponding block has not been transformed, ie the residual data represents a residual sample.

  As a background, Trans_quant_Bypass_flag indicates whether the corresponding block has been transformed and quantized. When Trans_quant_Bypass_flag is equal to 1, the corresponding block has not been transformed and quantized, ie the residual data represents a residual sample. In addition, Trans_quant_Bypass_flag is equal to 1 when the encoding is in lossless coding mode because the transform is used in regular coding mode but not in lossless coding mode

  Of course, in one example, at least one level encoding step, such as Greeter_than_1 and / or Greeter_than_2, may be selectively skipped for blocks of data to be encoded. In one example, checking Transform_skip_flag of a block of data is used to determine whether at least one level encoding step is skipped. In one example, at least one level encoding step may be skipped if a block of data is encoded using a lossless encoding mode. Of course, an increase in throughput can be realized when at least one level encoding stage is skipped.

  In one example, when a second different encoding technique is used, the electronic device 4121 may set a corresponding indicator, such as a flag associated with the second different encoding technique, to a value of 1 (this behavior). Of course, it may include changing the default value of the flag or leaving the flag at the default value depending on the priority of the design). However, in some examples, the decoding side checks the Transform_skip_flag and / or Trans_quant_Bypass_flag for the block and if either of the coding is equal to 1 if one of the coding is Transform_skip_flag and Trans_quant_Bypass_flag. Such explicit signaling is not necessary.

  FIG. 43 is a flowchart showing one configuration of a method for high-throughput residual coding on the decoding side.

  At block 4310, the electronic device 4122 obtains a bitstream. In block 4311, the electronic device 4122 restores a binary symbol from the obtained bitstream.

  At diamond 4312, the electronic device 4122 determines whether the binary symbol is decoded using high-throughput residual coding. In one example, the diamond portion 4312 determines whether the reconstructed binary symbol represents a transform coefficient, for example, determines whether Transform_skip_flag related to the reconstructed binary symbol is set, Trans_quant_bypass_flag is Determining whether it is set, and / or determining whether the bitstream has been encoded using a lossless encoding mode, and the like. If the condition is not satisfied at diamond 4312, at block 4313, electronic device 4122 obtains a block of TQC using the first encoding technique.

  If the condition is satisfied at diamond 4312, electronic device 4122 at block 4314 obtains a residual sample using a second different encoding technique. In block 4315, the electronic device 4122 stores the obtained block of TQC or the obtained residual sample in a memory device and / or recovers the video data.

  In one example, of the first and second encoding techniques, only the first encoding technique includes the Absolute-3 encoding technique. In one example, the second encoding technique includes an Absolute-1 or Absolute-2 encoding technique.

  In one example, the first encoding technique includes encoding the Greeter_than_1 and Greeter_than_2 flags, i.e., the GR1 encoding stage and the GR2 encoding stage, and the second encoding technique encodes any Greeter_than_1 flag and / or Greeter_than_2 flag. Ie, it does not include the GR1 encoding stage and / or does not include the GR2 encoding stage.

  In one example, the first encoding technique includes a Golomb-Rice (GR) encoding method for encoding an Absolute-3 value, and the second encoding technique uses an Absolute-1 or Absolute-2 value. Includes a G-R encoding method for encoding. In one example, the GR encoding method initially initializes the Rice parameter with a predefined value based on at least one selected from the group including block type, block size, and color information (luminance / color difference). Including the step of: In one example, the predefined rice parameters may depend on the texture (brightness / color difference). For example, the predefined Rice parameter value is 2 for the luminance block and 1 for the chrominance block. In one example, the same G-R encoding is used for Absolute-1 or Absolute-2 encoding and Absolute-3 encoding.

  In one example, the system obtains a bitstream, recovers a binary symbol from the obtained bitstream, and determines whether the binary symbol is decoded using a high-throughput residual coding mode. And a block of coefficients (TQC) transformed and quantized using the first coding technique in response to determining that the binary symbols are not decoded using the high throughput residual coding mode. And obtaining a residual sample using a second different coding technique in response to determining that the binary symbol is decoded using a high-throughput residual coding mode; The video data representing the obtained TQC block or the obtained residual sample, or the obtained TQC block or the obtained residual sample are stored in a memory device. Configured to perform the storing the scan, including a first electronic device of the decoder.

  In one example, of the first and second encoding techniques, only the first encoding technique includes Absolute-3 encoding. In one example, the second encoding technique includes Absolute-1 or Absolute-2 encoding.

  In one example, the first electronic device is responsive to determining whether the recovered binary symbol represents a transform coefficient and determining that the recovered binary symbol does not represent a transform coefficient. And using a second different encoding technique. In one example, the first electronic device determines whether a conversion skip flag or a transform quantization bypass flag associated with the recovered binary symbol is set, and the conversion skip flag or the transform quantization bypass flag is set. In response to determining that it is set, the second different encoding technique is used. In one example, the first electronic device determines whether a conversion skip flag or a transform quantization bypass flag associated with the recovered binary symbol is set, and the conversion skip flag or the transform quantization bypass flag is set. In response to determining that it is set, the second different encoding technique is used. In one example, in response to determining whether the bitstream was encoded using the lossless encoding mode and determining that the bitstream was encoded using the lossless encoding mode, the second different encoding Use technology.

  In one example, the first encoding technique includes encoding the Greater_than_1 flag and the Greator_than_2 flag, and the second encoding technique does not encode any of the Greater_than_1 flag and / or the Greater_than_2 flag.

  In one example, the bitstream originates from an encoder based on context adaptive binary arithmetic coding (CABAC).

  In one example, the first encoding technique includes a Golomb-Rice (GR) encoding method for encoding an Absolute-3 value, and the second encoding technique uses an Absolute-1 or Absolute-2 value. Includes a G-R encoding method for encoding.

  In one example, the CR encoding method initially initializes the Rice parameter with a predefined value based on at least one selected from the group including block type, block size, and color information (luminance / color difference). Including the step of: In one example, the electronic device 4121 may determine a predefined value for the bitstream or block based on the initialization information of the bitstream. The initialization information may include a rice parameter initialization value for the bitstream or the current block, i.e. per block. If the initialization information includes more than one Rice parameter initialization value for a block, the initialization information may further indicate criteria for using the first value or the second value, i.e., The value of may be 2 and 1, and the condition may be 2 for the luminance block and / or 1 for the chrominance block. In one example, the additional information may include additional syntax elements.

  In one example, the electronic device 4121 may determine a predefined value in response to determining the number of non-zero coefficients for the current block. For example, the electronic device 4121 uses the first predefined value if the number of non-zero coefficients is greater than a threshold, and if the number of non-zero coefficients is less than or equal to the threshold, May use a different second predefined value.

  The systems and apparatus described above may use a dedicated processor system, microcontroller, programmable logic device, microprocessor, or any combination thereof to perform some or all of the operations described herein. Good. Some of the operations described above may be implemented in software, and other operations may be implemented in hardware. One or more of the operations, processes and / or methods described herein may be performed by apparatus, devices and / or devices substantially similar to those described herein with reference to the illustrated drawings. It may be executed by the system.

  The processing device may execute instructions or “code” stored in the memory. The memory can also store data. The processing device may include, but is not limited to, an analog processor, digital processor, microprocessor, multi-core processor, processor array, or network processor. The processing device may be part of an integrated control system or system manager, or may be provided as a portable electronic device configured to interface with a network system locally or remotely via wireless transmission. Good.

  The processor memory may be integrated with the processing device, such as RAM or flash memory disposed in an integrated circuit microprocessor or the like. In other examples, the memory may include an independent device such as an external disk drive, a storage array, or a portable flash key fob. The memory and processing device may be operatively coupled together, such as by an I / O port or network connection, or may communicate with each other and the processing device may read a file stored in the memory. The associated memory may or may not be an intentional “read only” (ROM) with permission settings. Other examples of memory may include, but are not limited to, WORM, EPROM, EEPROM, flash, etc. that may be implemented in a solid state semiconductor device. Other memories may include moving parts, such as a conventional rotating disk drive. All of these memories may be “machine readable” or readable by a processing device.

  The operating instructions or commands may be implemented or embodied in the tangible form of stored computer software (also known as “computer programs” or “codes”). The program or code may be stored in digital memory and read by a processing device. "Computer-readable storage medium" (or alternatively "machine-readable storage medium") is stored as long as the memory can at least temporarily store digital information of the nature of a computer program or other data As long as the information can be “read” by a suitable processing device, it may include all of the aforementioned types of memory and future new technologies. The term “computer readable” may not be limited to the historical usage of “computer”, which implies a complete mainframe, minicomputer, desktop or laptop computer. Rather, “computer readable” may include storage media that may be readable by a processor, processing device or any computer system. Such a medium may be any available medium that may be locally and / or remotely accessible by a computer or processor, volatile and non-volatile media, removable and non-removable media, or any May be included.

  The program stored on the computer readable storage medium may include a computer program product. For example, a storage medium can be used as a convenient means for storing or carrying a computer program. For purposes of simplicity, operation may be described as various interconnected or coupled functional blocks or drawings. However, these functional blocks or drawings may be equally aggregated into a single logical device, program, or operation with unclear boundaries.

  Those skilled in the art will recognize that the concepts taught herein can be adapted to a particular application in many other ways. In particular, those skilled in the art will recognize that the illustrated embodiment is only one of many alternative implementations that will become apparent upon reading this disclosure.

  At some places in this specification, reference may be made to examples of “an”, “one”, “another”, or “some”. This reference does not necessarily mean that each such reference is to the same example (s) or that its features apply only to a single example.

Claims (16)

A system,
A first electronic device of a decoder, the first electronic device comprising:
Getting a bitstream,
Reconstructing binary symbols from the resulting bitstream;
Determining whether the binary symbol is decoded using a high throughput residual coding mode;
Responsive to determining that the binary symbols are not decoded using the high-throughput residual coding mode, a block of coefficients (TQC) transformed and quantized using a first coding technique is obtained. And
Responsive to determining that the binary symbol is decoded using the high throughput residual encoding mode, obtaining a residual sample using a second different encoding technique;
Storing the obtained TQC block or the obtained residual sample, or video data representing the obtained TQC block or the obtained residual sample in a memory device. And
Of the first and second encoding techniques, only the first encoding technique includes Absolute-3 encoding,
The second encoding technique includes either Absolute-1 or Absolute-2 encoding, and if both the Greeter_than_1 encoding step and the Greeter_than_2 encoding step are skipped, the Absolute-1 value is encoded, and if the Greeter_than_2 encoding is performed. A system in which the stages are skipped and the Absolute_than_1 encoding stage is encoded unless an Absolute-2 value is encoded . The first electronic device further includes
Determining whether the restored binary symbol represents a transform coefficient;
The method of claim 1, configured to perform the second different encoding technique in response to determining that the restored binary symbol does not represent the transform coefficient. system. The first electronic device further includes
Determining whether a transform skip flag or transform quantization bypass flag associated with the restored binary symbol is set;
In response to determining that the conversion skip flag or the transformed and quantized bypass flag is set, configured to perform a using said second different coding techniques, according to claim 2 System. The first electronic device further includes
Determining whether the bitstream was encoded using a lossless encoding mode;
In response to determining that the bit stream is encoded using the lossless coding mode, is configured to perform a using said second different coding techniques, according to claim 2 system.   2. The system of claim 1, wherein the first encoding technique includes encoding a Greeter_than_1 flag and a Greeter_than_2 flag, and wherein the second encoding technique does not encode any Greeter_than_1 flag and / or Greeter_than_2 flag. .   The system of claim 1, wherein the bitstream originates from an encoder based on context adaptive binary arithmetic coding (CABAC).   The first encoding technique includes a Golomb-Rice (GR) encoding method for encoding an Absolute-3 value, and the second encoding technique encodes an Absolute-1 or Absolute-2 value. The system of claim 1, comprising the G-R encoding method for converting. The GR encoding method initializes the rice parameters with a predefined value based on at least one selected from the group comprising block type, block size, and color information (luminance / color difference). The system of claim 7 , comprising steps. Obtaining a bitstream;
Reconstructing binary symbols from the obtained bitstream;
Determining whether the binary symbols are decoded using a high throughput residual coding mode;
Responsive to determining that the binary symbols are not decoded using the high-throughput residual coding mode, a block of coefficients (TQC) transformed and quantized using a first coding technique is obtained. Steps,
Responsive to determining that the binary symbol is decoded using the high throughput residual encoding mode, obtaining a residual sample using a second different encoding technique;
See containing and storing the video data representing the block or the resulting residue sample of the resulting TQC blocks or the resulting residue sample or the resulting TQC, the memory device,
Of the first and second encoding techniques, only the first encoding technique includes Absolute-3 encoding,
The second encoding technique includes either Absolute-1 or Absolute-2 encoding, and if both the Greeter_than_1 encoding step and the Greeter_than_2 encoding step are skipped, the Absolute-1 value is encoded, and if the Greeter_than_2 encoding is performed. The method wherein the stage is skipped and the Absolute_than_1 encoding stage is encoded unless an Absolute-2 value is encoded . Determining whether the restored binary symbol represents a transform coefficient;
10. The method of claim 9 , further comprising: using the second different encoding technique in response to determining that the restored binary symbol does not represent the transform coefficient. Determining whether a transform skip flag or transform quantization bypass flag associated with the restored binary symbol is set;
11. The method of claim 10 , further comprising: using the second different encoding technique in response to determining that the transform skip flag or the transform quantization bypass flag is set. Determining whether the bitstream has been encoded using a lossless encoding mode;
11. The method of claim 10 , further comprising: using the second different encoding technique in response to determining that the bitstream has been encoded using the lossless encoding mode. The method of claim 9 , wherein the first encoding technique includes encoding a Greeter_than_1 flag and a Greator_than_2 flag, and wherein the second encoding technique does not encode any of the Greater_than_1 flag and / or the Greater_than_2 flag. . The method of claim 9 , wherein the bitstream originates from an encoder based on context adaptive binary arithmetic coding (CABAC). The first encoding technique includes a Golomb-Rice (GR) encoding method for encoding an Absolute-3 value, and the second encoding technique encodes an Absolute-1 or Absolute-2 value. The method according to claim 9 , comprising the G-R encoding method for converting. The GR encoding method initializes the rice parameters with a predefined value based on at least one selected from the group comprising block type, block size, and color information (luminance / color difference). The method of claim 15 , comprising steps.