JP2007074648A - Cabac decoding apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a CABAC decoding apparatus for increasing an output data amount per unit time without including a plurality of decoder circuits. <P>SOLUTION: The present invention relates to an apparatus for decoding data encoded by context-based adaptive binary arithmetic coding (CABAC) and the apparatus comprises: an input selector 116 for acquiring a plurality of coded data that are encoded data; a context parameter table 105a, context parameter table 105b and context parameter table 105c corresponding to each of the plurality of coded data; and a decoding section 103 for decoding the plurality of coded data using the context parameter tables corresponding to each of the data. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a CABAC decoding apparatus that decodes data that has been arithmetically encoded by CABAC.

  Along with the development of digital technology, the technology for encoding video has also evolved and developed. However, the amount of data of video information, particularly moving image information, is very large. When a digital video encoded data is transferred by a medium such as a broadcast or a DVD, the transfer amount becomes very large. In particular, in high-definition broadcasting that has recently been put into practical use, the data amount is six times that of conventional SD (Standard Definition) video.

  With the development of digital video technology, in order to process the increasing amount of data, a technology for compressing data is used and developed for digital video data. The development has emerged as a compression technique specialized for video data that takes advantage of the characteristics of video data. In addition, with the improvement of information processing capabilities of computers and other devices, complex computations in compression techniques are possible, and the compression rate of video data has been greatly increased. For example, a compression technique employed in satellite and terrestrial digital high-definition broadcasting is a method called MPEG2, but AVC / H. H.264 is standardized by ITU-T, a telecommunications standardization department of the International Telecommunication Union (ITU).

  AVC / H. H.264 is a standard that realizes a compression rate about twice that of MPEG2. AVC / H. H.264 implements many compression technologies and combines them to achieve a high compression ratio. For this reason, the amount of calculation is also greatly increased.

  AVC / H. One compression technique implemented in H.264 is entropy coding (variable length coding). Two systems, CAVLC and CABAC, are prepared as entropy encoding systems. CAVLC is an abbreviation for Context-Adaptive Variable Length Coding (context-adaptive variable-length coding system). When encoding DCT coefficients, CAVLC uses a variable-length coding table for a run and a level that are consecutive zeros. In this method, encoding is performed from the direction opposite to the scanning direction.

  CABAC is an abbreviation for Context-based Adaptive Binary Arithmetic Coding (context-adaptive binary arithmetic coding system), and is a system that changes the appearance frequency of an encoding target that changes with time. Moreover, it is a method generally called arithmetic coding. In CABAC, in addition to normal arithmetic coding, a context index (ctxIdx) is attached to each code to be compressed, and the appearance frequency is changed for each ctxIdx.

  The encoding in CABAC is mainly divided into two processes. The first is a process called binarization that converts multi-value information to be encoded, called syntax, into binary data. The second is a process of calculating a context and performing arithmetic coding on binary data obtained by binarization.

  The process of decoding data encoded by CABAC (hereinafter referred to as “encoded data”) is also divided into two processes as in the case of encoding. This is a process of performing arithmetic decoding on the encoded data and outputting binary data, and a process of performing multi-value conversion for converting the binary data into Syntax.

  FIG. 8 is a diagram illustrating an example of a functional configuration of a conventional CABAC decoding apparatus. The CABAC decoding apparatus 200 shown in FIG. 8 is an apparatus that receives a bit stream and outputs binary data. Note that the input bit stream is processed in units of slices, which are a set of one or more macroblocks in one picture. A macroblock is image data composed of 16 × 16 pixels. Each bit of the output binary data is referred to as binVal.

  The CABAC decoding apparatus 200 includes a flip-flop circuit (FF) 201, an FF 202, a decoding unit 203, a ctxIdx calculation unit 204, and a context variable table 205.

  The outline of arithmetic decoding processing in the CABAC decoding apparatus 200 will be described below.

  The arithmetic decoding process is started by first obtaining ctxtIdx. ctxtIdx is obtained by calculation from the syntax of 1-bit binary data to be decoded and binIdx indicating the position of the binary data in the syntax. Also, these Syntax and binIdx are obtained by calculation from binVal, which is the binary data value of a macroblock that has already been decoded.

  These calculations are performed by the ctxIdx calculation unit 204. In the first calculation, the syntax given as the initial value and binidx = “0” are used.

  Next, using the obtained ctxtIdx, an occurrence probability pStateIdx currently assigned to the ctxtIdx and information valMPS representing a symbol with a high occurrence probability are extracted from the context variable table 205.

  The context variable table is a table in which ctxtIdx, pStateIdx, and valMPS are stored in association with each other.

  The read pStateIdx and valMPS are acquired by the decoding unit 203 via the FF 201. The decoding unit 203 further reads out codIRRange and codIOoffset, which are section information used in arithmetic coding, from the FF 202, and inputs pStateIdx and valMPS as inputs and performs an operation with codIRrange and codIOoffset. One bit of binary data is output by this calculation. The outputted 1 bit is binVal.

  At this time, codIRange and codIOoffset are updated and used for the next binVal operation. Also, according to the output binVal value, the occurrence probability pStateIdx and the information valMPS representing the symbol with a high occurrence probability are updated and written back to the context variable table 205.

  As described above, binVal, which is one bit of binary data, is output from the input bit stream by arithmetic decoding. Also, the next Syntax and binIdx values are determined by the output binVal value, and the decoding process for the next binVal output is started.

  FIG. 9 is an example of a time chart of the decoding process in the conventional CABAC decoding apparatus. Each of “An” (n = 0, 1, 2,...) In FIG. 9 is a symbol indicating the existence of various values related to the output of the (n + 1) th binVal. The horizontal axis represents time, and each of “T1, T2,...” Is a symbol indicating a cycle for each clock.

  For example, in cycle T1, there is a data input [A1] for outputting the second binVal, and pStateIdx and valMPS are values [A0] for calculating the first binVal. Also, the first binVal [A0] is calculated from the pStateIdx and valMPS. Although not represented on the time chart, the context variable table is updated according to the calculated binVal.

  In addition, Syntax and binIdx for outputting the next binVal are determined from the calculated binVal, and ctxIdx [A1] is calculated. Using the calculated ctxIdx [A1], pStateIdx [A1] and valMPS [A1] for outputting the second binVal are read from the context variable table. Those values read out are used in the next cycle T2.

  The above operation is repeated every clock while changing the data to be processed, and one binVal, which is one bit of binary data, is output every clock from the input bit stream.

Techniques for encoding and decoding such CABAC are also disclosed (for example, Patent Document 1).
JP 2004-135251 A

  As described above, in the CABAC decoding, the identification of the syntax and binIdx that is first performed to output one bit of binary data is determined by the value of one bit of the binary data output one time before. That is, unless the binary data output by the decoding process performed at that time is determined to be 0 or 1, the process for obtaining the next binary data cannot be performed.

  In addition, the values of pStateIdx and valMPS obtained from ctxIdx are changed according to the output binary data value and written back to the context variable table. Therefore, until these processes are completed, the next binary data is decoded by 1 bit. It is not possible to refer to the context variable table, which is an operation required for conversion.

  In other words, the circuit for decoding in the CABAC decoding device is a large loop circuit for each bit of binary data to be output, and the time taken for this loop is to output 1-bit output data from the CABAC decoding device. It will be the time required to get.

  Usually, when increasing the throughput of a circuit, a flip-flop circuit (FF) is inserted in the circuit to pipeline the process and increase the frequency. However, in the above-mentioned conventional CABAC decoding apparatus, 1 bit of binary data is output in one loop, so even if the loop is partitioned by FF to increase the frequency, the processing amount per clock Only decreases, and the throughput cannot be increased. Thereby, for example, the decoding process does not catch up with the reproduction of the image, and there is even a possibility that the reproduced video is interrupted.

  Thus, it is difficult for the conventional CABAC decoding device to increase the throughput of output data, that is, the number of binVal outputs per unit time. Of course, if a plurality of decoding circuits are prepared in parallel, the throughput can be improved, but in this case, for example, the manufacturing cost increases.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a CABAC decoding apparatus for increasing the amount of output data per unit time without having a plurality of decoding circuits.

  In order to solve the above problems and achieve the above object, a CABAC decoding apparatus according to the present invention is an apparatus for decoding data encoded by a context adaptive binary arithmetic coding system (CABAC), Acquisition means for acquiring a plurality of encoded data, which are encoded data, a plurality of context variable tables associated with each of the plurality of encoded data, and a plurality of the encoded data, respectively. And decoding means for decoding using the context variable table.

  Thus, the CABAC decoding apparatus of the present invention includes a plurality of context variable tables that acquire a plurality of encoded data and correspond to the plurality of encoded data. With this configuration, a plurality of pieces of encoded data can be decoded independently and in parallel, and as a result, the output amount per unit time of the decoded data can be improved.

  Further, in the CABAC decoding apparatus according to the present invention, the decoding means, and the plurality of context variable tables are used to pipeline each of the plurality of encoded data, and the number of the context variable tables is: The number of pipeline stages may be the same as the number of pipeline stages in the pipeline processing, and each of the plurality of encoded data is included in each of a series of different data strings. The number is the same as the number of pipeline stages, and the plurality of context variable tables are associated with each of the plurality of encoded data by being associated with each of the plurality of series of data strings. And the acquisition means has a predetermined order for each cycle among the series of data strings. By selecting one of said series of data strings, may acquire the encoded data included in said selected set of data strings.

  As described above, the CABAC decoding apparatus according to the present invention can improve the operating frequency by pipelining the decoding process, and thus can increase the amount of output data per unit time.

  Also, by setting the number of input data strings to be the same as the number of pipeline stages, it is possible to efficiently fill the gaps in the pipeline processing, and the processing efficiency is improved. Also, the encoded data to be subjected to the decoding process is input to the CABAC decoding apparatus by selecting a plurality of series of data strings including each of the plurality of encoded data in a predetermined order.

  For example, when the pipeline stage of the CABAC decoding apparatus is composed of three stages, three data string inputs A, B, and C are prepared. Three context variable tables corresponding to each data string, a, b, and c, are also prepared. When data string A is selected as an input to the CABAC decoding apparatus in cycle T0, data string B is selected in the next cycle T1, and data string C is selected in the next cycle T2. Further, in the next cycle T3, the data string A is selected again, and thereafter, the apparatus can be operated by changing the input of the data string every cycle of B → C → A → B →.

  The context variable tables a, b, and c are associated with the three data strings A, B, and C, respectively. Therefore, when the data string A is selected as an input, the context variable table a is referenced and updated. When the data string B is selected, the context variable table b is referenced. When the data string C is selected, the context variable table c is referenced. Updated. As a result, the context variable table corresponds to data input, and is referred to in order of a, b, and c for each cycle, and necessary information is read out. Also, it is updated in the order of reference.

  As described above, the CABAC decoding apparatus according to the present invention can efficiently perform the decoding process, and can increase the output data amount per unit time.

  Furthermore, the present invention can be realized as a method using the characteristic components of the CABAC decoding apparatus of the present invention as steps, or as a program including those steps, or a CD-ROM in which the program is stored. It can also be realized as a storage medium such as an integrated circuit. The program can also be distributed via a transmission medium such as a communication network.

  The present invention can provide a CABAC decoding apparatus for increasing the amount of output data per unit time without having a plurality of decoding circuits.

  Specifically, the CABAC decoding apparatus according to the present invention can acquire a plurality of pieces of encoded data and can process each of them independently. This makes it possible to process a plurality of encoded data in parallel. In addition, since the processing load per clock is reduced, the operating frequency of the circuit can be increased.

  For example, when a circuit is pipelined in order to improve the operating frequency, normally, one output bit can be obtained only for every clock corresponding to the number of stages of the pipeline, but by implementing the present invention, It is possible to obtain 1 output bit per clock.

  As a result, according to the present invention, it is possible to provide a CABAC decoding apparatus that can increase the amount of output data per unit time without having a plurality of decoding circuits.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  First, the configuration of the CABAC decoding apparatus according to the embodiment of the present invention will be described with reference to FIG. 1 and FIG.

  FIG. 1 is a functional block diagram showing a functional configuration of a CABAC decoding apparatus according to an embodiment of the present invention.

  The CABAC decoding apparatus 100 is an apparatus that performs decoding processing on three input slices in parallel and outputs binary data corresponding to each slice. Pipeline processing is divided into three pipeline stages.

  The CABAC decoding apparatus 100 inputs the FF 101, the FF 102, the decoding unit 103, the ctxIdx calculation unit 104, and three context variable tables (context variable table 105a, context variable table 105b, and context variable table 105c). An input selector 116 that selects a slice, a table selector 117 that reads pStateIdx and valMPS from the context variable table corresponding to the data to be processed at that time, and an FF (FF111 to FF115) for connecting each pipeline stage The output buffer 118 temporarily stores binVal obtained by the decoding unit 103.

  The decoding unit 103 is an example of a decoding unit in the CABAC decoding apparatus of the present invention, and the input selector 116 is an example of an acquisition unit. The ctxIdx calculation unit 104 is an example of a ctxIdx calculation unit in the CABAC decoding apparatus of the present invention, and the table selector is an example of a reading unit.

  In the CABAC decoding apparatus 100 of the present embodiment, the decoding processing method for one slice is the same as that of the conventional CABAC decoding apparatus 100. However, unlike the conventional CABAC decoding apparatus, it is an apparatus that pipelines three slices in parallel. For this reason, a context variable table is provided for handling each slice.

  In the present embodiment, each of the three input slices is set as sliceA, sliceB, and sliceC, and is selected by the input selector 116, whereby the slice data of the selected slice is input to the CABAC decoding apparatus 100.

  Further, the context variable table 105a, the context variable table 105b, and the context variable table 105c are context variable tables corresponding to the decoding processing of each slice. Note that these context variable tables are specifically held in one or more SRAMs (Static Random Access Memory).

  The ctxIdx calculation unit 104 is a processing unit that obtains a Syntax and a binIdx using the previously decoded binVal, and calculates ctxIdx from the two pieces of information.

  The decoding unit 103 is a processing unit that calculates binVal, which is 1-bit binary data, from the input slice data. Within the decryption unit 103 is a data table 1032 that is used to calculate binVal. In addition, it has a data table 1031 for determining how to update the context variable table according to the calculated binVal.

  Specifically, the data table 1031 is AVC / H. H.264 is a state transition table, and the data table 1032 is also an AVC / H. H.264 specification of rangeTabLPS dependent on pStateIdx and qCodIrangeIdx.

  The output buffer 118 is a buffer that accumulates binVal obtained by the decoding unit 103 for each slice data.

  FIG. 2 is a diagram illustrating an example of a data configuration of the context variable table.

  As illustrated in FIG. 2, the context variable table is a table for specifying pStateIdx and valMPS by ctxIdx. For example, when ctxIdx = “0” is output by the ctxIdx calculation unit 104, pStateIdx = “50” and valMPS = “0” are specified. The values of pStateIdx and valMPS are read by the table selector 117. Furthermore, it is acquired by the decoding unit 103 via the FF 101 and used to calculate binVal.

  Each of the context variable table 105a to the context variable table 105c has a data configuration as shown in FIG.

  Next, the operation of the CABAC decoding apparatus 100 will be described with reference to FIGS.

  FIG. 3 is a flowchart showing a flow of decoding processing in the CABAC decoding apparatus 100. The flow of processing in the CABAC decoding apparatus 100 from the input of certain slice data until the first binVal for that slice data is calculated and the processing for calculating the next binVal is performed will be described using FIG. .

  First, various initial values for obtaining binVal are set (S10). Specifically, the initial values of codIRange and codIOoffset are previously held in the FF 102 in advance. The initial value of Syntax is obtained from the slice coding conditions set by the slice header, PPS (picture parameter set), etc., and the initial value is given to ctxIdx. Also, this is the first decoding process, and binIdx indicating the position of binary data in the Syntax is set to “0”.

  FIG. 4 is a diagram illustrating an example of a data configuration of slice data. As shown in FIG. 4, the slice data has a header portion and a substance data portion, and slice setting data exists in the header portion. The initial value of Syntax is determined by slice setting data. The entity data portion includes one or more pieces of encoded data that is a target of decoding processing by the decoding unit 103.

  The slice data is an example of a series of data strings in the CABAC decoding apparatus according to the present invention.

  After setting the initial value, the ctxIdx calculation unit 104 calculates ctxIdx from the Syntax and binIdx (S11). For the first calculation, the above-mentioned initial value of Syntax and binIdx = “0” are used.

  When ctxIdx is calculated, pStateIdx and valMPS associated with the ctxIdx are obtained from the context variable table (S12). For example, ctxIdx obtained when processing slice data of sliceA is collated with the context variable table 105a, and pStateIdx and valMPS corresponding to the ctxIdx are obtained.

  The decryption unit 103 calculates binVal using the obtained pStateIdx and valMPS, and codIRange and codIOoffset read from the FF 102 (S13).

  Also, codIRrange (new_codIRrange) and codIOoffset (new_codIOoffset) for calculating the next binVal for the same slice are calculated (S14).

  Also, a new pStateIdx (new_pStateIdx) and a new valMPS (new_valMPS) are calculated from the calculated binVal and the data table 1031. The corresponding context variable table is updated with the calculated values (S15).

  The ctxIdx calculation unit 104 obtains the Syntax and binIdx necessary for calculating the next ctxIdx for the same slice based on the binVal calculated by the decoding unit 103 (S16). When Syntax and binIdx are obtained, the process returns to the calculation of ctxIdx (S11) and the above operation is repeated.

  FIG. 5 is an example of a time chart of the decoding process in the CABAC decoding apparatus 100. In FIG. 4, “An” (n = 0, 1, 2,...) Means n + 1th bit of binary data 1 bit obtained by decoding sliceA slice data, that is, n + 1th binVal. Is a symbol indicating the existence of various values related to the output of. The horizontal axis represents time, and each of “T1, T2,...” Is a symbol indicating a cycle for each clock.

  For example, the fact that pStateIdx and valMPS are “A1” in cycle T4 means that processing using pStateIdx and valMPS to obtain the second binVal from the slice data of slice A is performed in cycle T4. Show. The same applies to “Bn” and “Cn”.

  In the time chart of FIG. 5, the flow of processing for obtaining the second binVal [A1] from the input data of sliceA will be described for each cycle.

  In cycle T2, the next ctxIdx [A1] is obtained using binVal [A0] already obtained.

  In cycle T3, the values of pStateIdx [A1] and valMPS [A1] are obtained using ctxIdx [A1] obtained in cycle T2. At the same time, the slice A data [A1] is selected as input data. Also, codIRrange [A1] and codIOoffset [A1] for obtaining binVal [A1] are obtained from the values of codIRRange [A0] and input data [A1] used to obtain binVal [A0].

  In cycle T4, binVal [A1] is obtained using pStateIdx [A1], valMPS [A1], codIRange [A1] and codIOoffset [A1].

  Although not expressed in the time chart, new values new_pStateIdx and new_valMPS are obtained by changing the values of pStateIdx [A1] and valMPS [A1] according to the value of binVal [A1]. It is done.

  In cycle T5, binVal [A1] is output to the output buffer 118. At this time, the context variable table 105a is updated by new_pStateIdx and new_valMPS, and processing for outputting binVal [A2] is started from the next cycle T6.

  Thus, data [A1] for obtaining binVal [A1] is input in cycle T3, and binVal [A1] is output in cycle T5. That is, when focusing on only slice A, binVal is output at a timing of once every three clocks.

  However, as described above, the CABAC decoding apparatus 100 performs decoding processing on three slice data in parallel at three pipeline stages. In the present embodiment, input data is selected in the order of sliceA, sliceB, and sliceC, and processed in parallel. As a result of processing, the binVal1 bit of each slice is output in the same order as the input every clock.

  For example, the contents of processing performed in parallel at each pipeline stage in cycle T6 of the time chart shown in FIG. 5 are as follows.

<Pipeline stage 0>
The decoding unit 103 uses binVal [C1] and codIOOffset [C1] set in the FF 102 and pStateIdx [C1] and valMPS [C1] obtained from the context variable table 105c in the pipeline stage 2. C1] and new_codIRrange [C2] and new_codIOOffset [C2].

  binVal [C1] is set in FF111, and new_codIRrange [C2] and new_codIOOffset [C2] are set in FF113.

  Also, pStateIdx [C1] and valMPS [C1] are changed according to the values obtained using the obtained binVal [C1] and the data table 1031. The changed values are set in the FF 112 as new_pStateIdx and new_valMPS for updating the context variable table 105c.

<Pipeline stage 1>
In pipeline stage 0, binVal [B1] obtained by the decoding unit 103 and set in the FF 111 is output to the output buffer 118. Also, using binVal [B1], the ctxIdx calculation unit 104 calculates the next ctxIdx [B2] and sets it in the FF 115.

  Also, new_pStateIdx and new_valMPS obtained by the decoding unit 103 in the pipeline stage 0 and set in the FF 112 are written into the context variable table 105b corresponding to sliceB. That is, the context variable table 105b is updated.

<Pipeline stage 2>
CtxIdx [A2] and valMPS [A2] corresponding to ctxIdx [A2] are referred to the context variable table 105a using ctxIdx [A2] obtained by the ctxIdx calculation unit 104 and set in the FF 115 in the pipeline stage 1 Is required.

  The obtained pStateIdx [A2] and valMPS [A2] are read by the table selector 117 and set in the FF101. Also, codIRrange [A2] and codIOoffset [A2] are created and set in FF102 by new_codIRrange [A2] and new_codIOOffset [A2] and input [A2] output from FF114.

  As described above, the CABAC decoding apparatus 100 according to the embodiment of the present invention improves the operating frequency by pipelining the decoding circuit.

  Here, in the CABAC decoding process, in order to obtain one binVal, that is, one bit of binary data, the previous one bit needs to be calculated. Therefore, even if an FF is inserted in the circuit and the operating frequency is simply improved, the output amount of the processing result per unit time cannot be increased.

  For example, as shown in FIG. 6, it is assumed that a plurality of FFs are inserted into a conventional decoding circuit to improve the operating frequency.

  FIG. 6 is a diagram showing an example in which a plurality of FFs are inserted into a conventional CABAC decoding device to improve the operating frequency.

  Similar to the CABAC decoding apparatus 100 according to the embodiment of the present invention, the CABAC decoding apparatus 150 shown in FIG. 6 includes a plurality of FFs and can perform pipeline processing including three pipeline stages. As a result, the processing load for each clock can be reduced and the operating frequency can be improved.

  FIG. 7 is an example of a time chart of the decoding process in the CABAC decoding apparatus 150.

  As shown in FIG. 7, the processing cycle speed is improved by improving the operating frequency. However, even in this case, a 1-bit output result is obtained every time the decoding processing loop is completed. Absent.

  For example, even if a conventional circuit operating at 100 MHz is divided into three stages and operated at 300 MHz, which is three times as much, a 1-bit output can be obtained only once every three cycles. That is, only the same output rate as that of the conventional circuit can be obtained.

  However, the CABAC decoding apparatus 100 according to the embodiment of the present invention not only simply inserts FFs and divides the processing into a plurality of stages to improve the operating frequency, but also inputs three slice data in a certain order. By capturing as data, the three pipeline stages can be operated most efficiently.

  Specifically, since each of the three slice data can be processed independently, the processing of other slice data can proceed regardless of the processing of certain slice data. That is, as in the present embodiment, the three pipeline stages can be efficiently operated without gaps by fetching the data of sliceA, sliceB, and sliceC in the order of A, B, and C.

  In addition, in order to process three slice data in parallel, three context conversion tables corresponding to decoding processing of each slice data are provided. These three context conversion tables exist independently of each other, and are updated and referred to according to the processing result of each slice data.

  As described above, the CABAC decoding apparatus 100 according to the embodiment of the present invention captures three slice data in a certain order in order to improve the operating frequency by pipeline processing and efficiently perform pipeline processing. . Since each of the acquired three slice data is processed in parallel by the three pipeline stages, as a result, the output data amount per unit time can be increased.

  In the present embodiment, it is assumed that the number of stages in the pipeline processing of the CABAC decoding apparatus 100 is 3, and the input slice data is 3. However, the number of pipeline stages and the number of input slice data need not be three.

  For example, the number may be four, the number of pipeline stages may be three, and the number of input slice data may be five. In other words, if the number of input slice data is larger than the number of pipeline stages, the processing is performed efficiently without causing a gap in pipeline processing.

  Further, for example, in the CABAC decoding apparatus according to the present embodiment, when sliceD is further subjected to decoding processing, the order in which the input selector 116 selects input data may not be the order in which sliceA to sliceD are repeated. Good. For example, after selecting sliceA, sliceB, sliceC, and sliceD, you may select sliceB and sliceA.

  That is, when the number of pipeline stages is 3, one binVal is output every three cycles. Therefore, if a certain slice is selected at an interval of 3 cycles or more, a new binVal can be calculated and output based on the calculated binVal. Accordingly, binVal can be calculated efficiently if the interval at which a slice is selected is the number of cycles equal to or greater than the number of pipeline stages.

  For example, when the number of pipeline stages is 3 and the number of input slice data is 2, a gap is generated in the pipeline processing. However, the amount of output data per unit time is larger than at least the conventional CABAC decoding device.

  When a plurality of processing steps are performed in order to obtain one output data, a plurality of processing steps are completely performed in series if there is a portion where processing is performed in parallel in the time series even a little. This is because the time required for processing is reduced compared to the case.

  In other words, if the number of input slice data is plural for a plurality of pipeline stages, at least the amount of output data per unit time can be increased as compared with the conventional CABAC decoding apparatus.

  As shown in FIG. 1, the CABAC decoding apparatus 100 includes an output buffer 118 that accumulates binVal obtained by the decoding process, but may include a plurality of output buffers. For example, when three slice data are targeted for parallel processing as in the present embodiment, three output buffers may be provided so as to correspond to each slice data.

  The present invention can be used in a decoding apparatus that decodes data encoded by CABAC. Specifically, AVC / H. H.264 is optimal as a decoding device for reproducing high-definition video encoded by a data compression technique defined in H.264.

It is a functional block diagram which shows the functional structure of the CABAC decoding apparatus of embodiment of this invention. It is a figure which shows an example of a data structure of a context variable table. It is a flowchart which shows the flow of the decoding process of the CABAC decoding apparatus of embodiment of this invention. It is a figure which shows an example of a data structure of slice data. It is an example of the time chart of the decoding process in the CABAC decoding apparatus of embodiment of this invention. It is a figure which shows an example made into the structure which inserts several FF in the conventional CABAC decoding apparatus and makes it possible to improve an operating frequency. 7 is an example of a time chart of a decoding process in the CABAC decoding apparatus shown in FIG. 6. It is a figure which shows an example of a functional structure of the conventional CABAC decoding apparatus. It is an example of the time chart of the decoding process in the conventional CABAC decoding apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 CABAC decoding apparatus 103 Decoding part 104 ctxIdx calculation part 105a, 105b, 105c Context variable table 101, 102, 111, 112, 113, 114, 115 Flip-flop circuit 116 Input selector 117 Table selector 118 Output buffer 1031, 1032 Data table

Claims (8)

An apparatus for decoding data encoded by a context adaptive binary arithmetic coding scheme (CABAC),
Obtaining means for obtaining a plurality of pieces of encoded data that is encoded data;
A plurality of context variable tables associated with each of the plurality of encoded data;
A CABAC decoding device comprising: decoding means for decoding a plurality of encoded data using the context variable table associated with each of the encoded data. Each of the plurality of encoded data is included in each of a series of different data strings,
The plurality of context variable tables are associated with each of the plurality of encoded data by being associated with each of the plurality of the series of data strings,
The acquisition unit acquires encoded data included in the selected series of data strings by selecting the series of data strings in a predetermined order from a plurality of the series of data strings. Item 5. The CABAC decoding device according to Item 1. The decoding means outputs binary data obtained by decoding the encoded data bit by bit,
The CABAC decoding device further includes:
CtxIdx calculating means for calculating a context index (ctxIdx) that is information about the type of the next bit, which is the next 1 bit obtained from the encoded data, based on 1 bit output from the decoding means;
With reference to the context variable table corresponding to the encoded data, read out probability information which is information about the probability of whether the next bit corresponding to ctxIdx calculated by the ctxIdx calculating means is 0 or 1 Reading means for passing to the decoding means,
The CABAC decoding apparatus according to claim 1, wherein at least two of the decoding unit, the ctxIdx calculating unit, and the reading unit perform processing on different encoded data in parallel. The decoding means and the plurality of context variable tables are used to pipeline each of the plurality of encoded data,
The CABAC decoding apparatus according to claim 1, wherein the number of the context variable tables is the same as the number of pipeline stages in the pipeline processing. Each of the plurality of encoded data is included in each of a series of different data strings,
The number of the plurality of series of data strings is the same as the number of pipeline stages,
The plurality of context variable tables are associated with each of the plurality of encoded data by being associated with each of the plurality of the series of data strings,
The acquisition unit selects one of the series of data strings in a predetermined order for each cycle from the plurality of series of data strings, thereby including encoded data included in the selected series of data strings. The CABAC decoding apparatus according to claim 4. The decoding means outputs binary data obtained by decoding the encoded data, one bit per cycle,
The CABAC decoding device further includes:
The CABAC decoding apparatus according to claim 4, further comprising an output buffer that accumulates one bit of the binary data output by the decoding means for each encoded data before being decoded. A method for decoding data encoded by a context adaptive binary arithmetic coding scheme (CABAC),
An acquisition step of acquiring a plurality of encoded data that are encoded data;
A decoding method comprising: a decoding step of decoding the plurality of encoded data using a plurality of context variable tables associated with each of the plurality of encoded data. An integrated circuit for decoding data encoded by a context adaptive binary arithmetic coding scheme (CABAC),
Obtaining means for obtaining a plurality of pieces of encoded data that is encoded data;
A plurality of context variable tables associated with each of the plurality of encoded data;
An integrated circuit comprising: decoding means for decoding a plurality of the encoded data using the context variable table associated with each of the encoded data.

Images