JP4241417B2 - Arithmetic decoding device and arithmetic decoding program - Google Patents

Description

  The present invention relates to an arithmetic decoding device and an arithmetic decoding program used in decoding of images, video, audio, and the like using arithmetic decoding.

  Conventionally, in moving picture coding represented by MPEG, VLC (Variable Length Code), which is variable length coding, is used as entropy coding used when a bitstream is output. Since VLC is performed based on a Huffman coding method, it is known that several problems such as causing an increase in the Huffman table (Non-Patent Document 1) are caused when an attempt is made to improve the coding efficiency. Therefore, in recent years, instead of VLC, JBIG (Non-Patent Document 2) uses QM-Coder, JPEG2000 (Non-Patent Document 3) uses EBCOT, and ITU-T H.264 (Non-Patent Document 4) (hereinafter referred to as AVC) uses VLC. In addition, various arithmetic coding / decoding methods such as CABAC (Context-based Adaptive Binary Arithmetic Coding) are adopted.

  The concept of alias coding / decoding, which is basic arithmetic coding and decoding, is shown in FIG. In arithmetic coding and decoding, in general, a real line with a lower limit value of 0 or more and an upper limit value of less than 1 is considered. This interval is the initial range.

  In arithmetic coding, a real line is divided based on the occurrence probability of a dominant symbol (hereinafter referred to as MPS) and an inferior symbol (hereinafter referred to as LPS). Let the divided areas be rMPS and rLPS, respectively. Thereafter, a region to which a symbol to be encoded output belongs is selected. Here, it is assumed that rMPS is selected. Thereafter, the selected rMPS is set as the next range, and this range is similarly divided into rMPS and rLPS, and an area to be encoded is selected. This operation is repeated until there are no encoding output targets, and the value that falls within the range of the range that is finally specified is output as the output value value, thereby completing the encoding. Normally, the lower limit value of the range is output as the output value value.

  Also, in arithmetic decoding, if the value to decode the output value value output in arithmetic encoding is the input value value, the real line is divided based on the occurrence probability of MPS and LPS as in arithmetic encoding. To do. Thereafter, an area including the input value value is specified. Here, rMPS is selected, and a symbol corresponding to this area is output as a decoded symbol. By repeating this operation, a necessary decoded symbol string can be obtained.

  The basic arithmetic coding / decoding described above is not practical because the range of the section and the value that is the input / output value are real values, and it is not practical due to the problem of calculation accuracy. Therefore, the real line is an integer interval having a finite bit width, and the problem is solved by introducing a normalization process to the range that is the width of the specified interval and the value that is the lower limit value of the specified interval. The normalization process is performed by a left shift operation, and the calculation accuracy of range and value is ensured by this normalization process.

  Arithmetic encoding / decoding such as QM-Coder, EBCOT, CABAC and the like that are currently generally used is also performed based on the subtraction shift type adaptive arithmetic encoding / decoding.

  CABAC uses a plurality of context variable information to improve the encoding efficiency. In order to arithmetically decode the bit stream encoded in this way, the decoder is as shown in FIG. 8 and FIG. (Details will be described later).

  The context variable information includes state information (hereinafter referred to as pStateIdx) that is a probability state used when performing arithmetic coding, and dominant symbol information (hereinafter referred to as valMPS). In order to specify a plurality of pieces of context variable information, a context index (hereinafter referred to as ctxIdx) is assigned to the structure of the context variable information. Here, the dominant symbol represents which of the 0 and 1 binary symbols encoded up to the present when a certain symbol sequence is arithmetically encoded.

  In arithmetic coding, generally, an input symbol string is coded by specifying a position from 0 to 1 and a section width using the occurrence probability of each symbol. When the occurrence probability of each symbol is known accurately in advance, encoding can be performed in a state closest to the entropy of the input symbol string. However, when the input symbol string is input in real time, it is difficult to accurately specify the occurrence probability of each input symbol. The greater the difference between the occurrence probability of each identified input symbol and the actual occurrence probability, the lower the coding efficiency.

  Performing arithmetic coding in consideration of performing real-time arithmetic coding on input symbol sequences that consist of a series of multiple symbol sequences, or on changes in the occurrence probability of each symbol in the input symbol sequence In this case, when all the input symbol sequences are encoded using the predetermined occurrence probability of each input symbol, the encoding efficiency is not matched with the actual occurrence probability of the input symbol sequence. Will fall.

  Even when the arithmetic coding device is an adaptive device that can adaptively update the occurrence probability according to the input symbol, for example, as shown in FIG. When an input symbol sequence consisting of a combination of symbol sequences that is clearly different from the sequence C is encoded, it takes a certain amount of time for the occurrence probability estimated from the input symbols to converge. When performing, mismatch of occurrence probability arises, and as a result, encoding efficiency will fall.

  Therefore, as shown in FIG. 10, when it is known in advance that the input symbol string is composed of a plurality of symbol sequences, a context corresponding to each sequence, that is, an optimal occurrence probability is prepared. In addition, it is possible to reduce the occurrence probability mismatch by switching to the corresponding context each time the symbol sequence is switched.

  Context switching is generally controlled according to a predetermined rule. In CABAC, at least one context variable information corresponding to each syntax information is specified by following the syntax, and appropriate context variable information is selected from at least one context variable information according to the internal state. Encoding efficiency is improved by such fine context switching control.

  Note that the syntax refers to the decoding procedure itself, and defines in what order the syntax elements to be decoded are encoded and decoded, and the syntax elements constitute the syntax. Examples of each element that refers to each element and constitutes syntax include vector information, information on DCT coefficients obtained by DCT, flag information for performing various mode determinations, and the like. Hereinafter, the syntax element is treated as syntax information.

  FIG. 8 is a functional block diagram of a decoder 101 that performs a process of decoding a bitstream and outputting syntax information in general AVC.

  As shown in FIG. 8, the decoder 101 includes a bit stream acquisition unit 102, a decoding position detection unit 103, a bit stream buffer 104, a decoding position management unit 105, a syntax information acquisition unit 106, a syntax information storage unit 107, and an arithmetic unit. The decoding unit 108 is configured.

  The bit stream acquisition unit 102 has a function of acquiring an encoded bit stream.

  The decoding position detection unit 103 searches the bit stream acquired by the bit stream acquisition unit 102 based on the start code obtained from the decoding position management unit 105, and detects a position in the bit stream that matches the start code. Has a function of specifying the start position of decoding. The decoding position detection unit 103 has a function of specifying a decoding unit by specifying the next decoding position. Here, a better configuration is achieved by enabling a bit string that can decode at least one slice to be specified as a decoding unit. A bit string in the bit stream specified by the decoding position detection unit 103 is stored in the bit stream buffer 104.

  The bit stream buffer 104 stores the bit string specified by the decoding position detection unit 103.

  The decoding position management unit 105 has a function of managing the decoding position in the input bitstream acquired by the bitstream acquisition unit 102. In addition, the decoding position management unit 105 has a function of managing the decoding position in the bit string stored in the bit stream buffer, and designates a start code to the decoding position detection unit 103 as necessary to generate a bit stream. The bit string in the buffer 104 is updated. Further, the decoding position management unit 105 has a function of notifying the syntax information acquisition unit 106 and the arithmetic decoding unit 108 of the decoding position information in the bit stream buffer 104.

  The syntax information acquisition unit 106 is a header positioned higher than the bit sequence encoded by the arithmetic decoding unit 108 from the decoding location information obtained from the decoding location management unit 105 and the bit sequence stored in the bit stream buffer 104. It has a function of decoding information, and then has a function of notifying syntax information storage section 107 of syntax information obtained.

  The syntax information storage unit 107 has a function of storing various pieces of syntax information obtained from the syntax information acquisition unit 106 and notifying the arithmetic decoding unit 108 of the syntax information as necessary.

  The arithmetic decoding unit 108 performs arithmetic decoding from the position of the bit string specified by the decoding position information from the decoding position information obtained from the decoding position management unit 105 and the bit string stored in the bit stream buffer 104, and the decoding result is obtained. It has a function of outputting as output syntax information.

  FIG. 11 is a flowchart showing the operation of the decoder 101 shown in FIG. The operation of the decoder 101 will be described below with reference to FIG.

  When the decoder 101 starts a bitstream decoding process, the bitstream acquisition unit 102 acquires an input bitstream to be decoded (step S101).

  The decoding position detection unit 103 acquires a start code necessary for specifying the decoding position from the decoding position management unit 105 (step S102), searches the bit stream acquired by the bit stream acquisition unit 102, and specifies a bit string. .

  Next, the decoding position detection unit 103 stores the bit string specified in step S102 in the bit stream buffer 104 (step S103).

  The syntax information acquisition unit 106 performs decoding based on the bit string stored in the bit stream buffer 104 and the decoding position information obtained from the decoding position management unit 105 to acquire syntax information (step S104). The syntax information is stored in the syntax information storage unit 116 (step S105).

  The arithmetic decoding unit 108 acquires a bit string from the bit stream buffer 104 and acquires decoding position information of the bit string from the decoding position management unit 105. Further, the arithmetic decoding unit 108 obtains necessary syntax information from the syntax information storage unit 107 in order to determine how to interpret and decode the bit at the position on the bit string specified from the decoding position information. get. Based on these pieces of information, the arithmetic decoding unit 108 performs arithmetic decoding (step S106), sequentially outputs output syntax information that is a decoding result, and the decoder ends the decoding process of the bitstream.

  FIG. 9 is a functional block diagram showing the arithmetic decoding unit 108 shown in FIG. 8 in more detail.

  The arithmetic decoding unit 108 includes a state transition management unit 111, a context initialization unit 112, a context initialization information storage unit 113, a context variable information storage unit 114, a bit string count unit 115, a bit string information management unit 116, a context specification unit 117, 2 The value arithmetic decoding unit 118, the decoded bit acquisition unit 119, and the bit string comparison unit 120 are configured.

  The state transition management unit 111 has a function of managing the internal state in the arithmetic decoding unit 108 so that syntax information that is currently being decoded can be correctly decoded according to the syntax. The state transition management unit 111 performs an initialization request to the context initialization unit 112, control of the bit string count unit 115, control of the bit string information management unit 116, control of the context identification unit 117, and decoding to the binary arithmetic decoding unit 118. The internal state is managed by making a start request.

  The context initialization unit 112 has a function of performing initialization processing of context variable information used when performing arithmetic decoding in response to an initialization request from the state transition management unit 111. The context initialization unit 112 initializes context variable information using a predetermined constant stored in the context initialization information storage unit 113 and the decoded syntax information. Decoded syntax information required for CABAC initialization is SliceQPY and slice_type and cabac_init_idc included in slice_header () obtained from pic_init_qp_minus26 included in pic_parameter_set_rbsp () and slice_qp_delta included in slice_header (). Detailed initialization of SliceQPY and context variable information will be described later with reference to FIG. Initialization of context variable information is generally performed at the head when decoding a decoding unit. In CABAC, since the decoding unit is a slice, context variable information is initialized at the head every time a slice is decoded. .

  The context initialization information storage unit 113 stores predetermined constants used when the context initialization unit 112 initializes context variable information. FIG. 12 is an excerpt of a part of a table in which predetermined constants used in CABAC are described as an example of stored context initialization information. In CABAC, constants m and n are identified from a table of context initialization information as shown in FIG. 12, using slice_type obtained from decrypted syntax information, cabac_init_idc, and ctxIdx for identifying each context. Used to initialize context variable information.

  The context variable information storage unit 114 stores the context variable information obtained by the initialization process of the context initialization unit 112. The context variable information stored in the context variable information storage unit 114 is changed according to the decoding result of the binary arithmetic decoding unit 118. In the configuration of FIG. 9, the context variable information is updated via the context specifying unit 117. In general, the binary arithmetic decoding unit 118 can directly change the contents of the context variable information storage unit 114. It is also possible to take

  The bit string count unit 115 has a function of counting which position in the bit string the bit currently being decoded corresponds to. This is because the occurrence probability of a bit differs depending on the bit position to be decoded, and therefore context variable information corresponding to the bit position needs to be associated.

  The bit string information management unit 116 has a function of converting information that becomes syntax information candidates to be decoded into information expressed in bit strings and performing association. This correspondence is used by the bit string comparison unit 120.

  The context specifying unit 117 obtains, from the bit string count unit 115, information regarding the type of syntax information that is to be decoded currently obtained from the state transition management unit 111, and ctxIdx for specifying the context variable information of the bit that is to be decoded. And a function of specifying based on the bit position in the currently decoded bit string and the decoded syntax information.

  The binary arithmetic decoding unit 118 has a function of performing arithmetic decoding using the context variable information specified by the context specifying unit 117. The context variable information whose state has been changed by arithmetic decoding is sequentially stored in the context variable information storage unit 114. In FIG. 9, this is performed via the context specifying unit 117. The binary arithmetic decoding unit 118 supplements necessary bits from the decoded bit acquisition unit 119 when it is necessary to replenish bits from the bitstream by normalization during arithmetic decoding.

  The decoded bit acquisition unit 119 has a function of acquiring a bit stream and decoding position information in response to a bit request from the binary arithmetic decoding unit 118.

  The bit string comparison unit 120 has a function of holding the output result of the binary arithmetic decoding unit 118 and comparing with the bit string information prepared by the bit string information management unit 116. If there is an equal bit string information, the bit string comparison unit 120 acquires syntax information corresponding to the equal bit string information, outputs the syntax information as output syntax information, and holds the held binary arithmetic decoding unit 118. Is prepared for decoding of the next syntax information. Further, the bit string comparison unit 120 notifies the state transition management unit 111 of output syntax information as necessary.

  FIG. 13 is a flowchart showing a decoding operation for one slice, which is a CABAC decoding unit. The operation of the arithmetic decoding unit 108 will be described below using FIG.

  When CABAC decoding is started, the state transition management unit 111 determines whether the syntax information to be decoded is the first syntax information in the slice (step S111). In the case of the first syntax information, an initialization request is made to the context initialization unit 112, and an initialization request is made to the binary arithmetic decoding unit 118. If it is not the first syntax information, the process proceeds to step S114.

  When the initialization request is received from the state transition management unit 111, the context initialization unit 112 initializes context variable information (step S112). The initialization of the context variable information will be described in detail later with reference to FIG.

  When the initialization request is received from the state transition management unit 111, the binary arithmetic decoding unit 118 initializes the range to a predetermined value and initializes the value with the bit acquired from the decoded bit acquisition unit 119 (step S113). ).

  Next, when the state transition management unit 111 requests the bit string information management unit 116 to convert information associated with syntax information candidates to be currently decoded into information represented by bit strings, the state transition management unit 111 is based on this request. The bit string information management unit 116 creates bit string information (step S114).

  Further, the bit string count unit 115 initializes the bit string counter binIdx in the bit string count unit 115 to −1 (step S115).

  Next, the bit string count unit 115 increments the bit string counter binIdx in the bit string count unit 115 by 1 (step S116).

  Next, when the state transition management unit 111 requests the context determination unit 117 to specify ctxIdx for specifying the context variable information used in decoding, the context determination unit 117 determines the bit that is currently being decoded. CtxIdx for specifying the context variable information, information on the type of syntax information to be decoded currently obtained from the state transition management unit 111, the bit position in the bit string to be decoded currently obtained from the bit string count unit 115, Identification is performed using the decoded syntax information (step S117).

  When the state transition management unit 111 makes a decoding request to the binary arithmetic decoding unit 118, the binary arithmetic decoding unit 118 acquires context variable information obtained based on the ctxIdx specified from the context specifying unit 117, Decoding is performed (step S118). When bit supplementation from the bitstream is necessary due to normalization at the time of arithmetic decoding, necessary bits are supplemented from the decoded bit acquisition unit 119. When the decoding is completed, the binary arithmetic decoding unit 118 notifies the bit string comparison unit 120 of the decoding result.

  Next, the bit string comparison unit 120 compares the output result from the binary arithmetic decoding unit 118 held by the bit string comparison unit 120 with the bit string information stored in the bit string information management unit 116 (step S119). If there is no equal bit string information, the process proceeds to step S116 and the process is continued. If there is an equal bit string information, the bit string comparison unit 120 acquires syntax information corresponding to the equal bit string information, outputs the syntax information as output syntax information, and holds the held binary arithmetic decoding unit 118. Is prepared for decoding of the next syntax information. Further, the bit string comparison unit 120 notifies the state transition management unit 111 of output syntax information as necessary.

  Next, the state transition management unit 111 determines whether or not the next syntax information to be decoded that is currently the decoding target is mb_type and the decoding result obtained in the process of step S118 is I_PCM. To do. If this condition is satisfied, the process proceeds to step S121. If the condition is not satisfied, the CABAC decoding is terminated.

  When the initialization request is received from the state transition management unit 111, the binary arithmetic decoding unit 118 initializes the range to a predetermined value and initializes the value with the bit acquired from the decoded bit acquisition unit 119 (step S121). ). When the initialization is completed, a series of CABAC decoding processes are completed.

  FIG. 14 is a flowchart illustrating in more detail the initialization of the context variable information (step S112) in FIG. The context variable information initialization process will be further described below with reference to FIG.

  When the initialization of the context variable information is started, first, in response to the initialization request from the state transition management unit 111, the context initialization unit 112 selects pic_init_qp_minus26, slice_qp_delta, slice_qp_delta of the slice currently being decoded from the decoded syntax information Slice_type and cabac_init_idc are acquired (step S131). Here, it is assumed that the context initialization unit 112 has acquired pic_init_qp_minus26 = 4, slice_qp_delta = 5, slice_type = 0 (P slice), and cabac_init_idc = 1.

  Next, the context initialization unit 112 calculates SliceQPY according to the following mathematical formula (step S132).

[Equation 1]
SliceQPY = 26 + pic_init_qp_minus26 + slice_qp_delta
= 26 + 4 + 5
= 35
Next, the context initialization unit 112 identifies ctxIdx that needs to be initialized from the decoded context variable information slice_type and cabac_init_idc acquired in step S131, and sets m and n corresponding to the ctxIdx to the context initialization information storage unit 113 (step S133). Assuming that ctxIdx requiring initialization is ctxIdx = 70, the context initialization unit 112 acquires m = 13 and n = 15 according to the context initialization information table shown in FIG.

  Next, the context initialization unit 112 calculates a temporary variable preCtxState according to the following formula using SliceQPY calculated in step S132 and m and n acquired in step S133 (step S134).

[Equation 2]
preCtxState = ((m × SliceQPY) >> 4) + n
= (13 × 35) >> 4 + 15
= 28 + 15
= 43
Here, “>> 4” in the equation indicates that a 4-bit right shift operation is performed.

  Next, the context initialization unit 112 determines whether or not the temporary variable preCtxState is smaller than 1 (step S135).

  When the temporary variable preCtxState is smaller than 1, the context initialization unit 112 sets 1 to the temporary variable preCtxState (step S136).

  When the temporary variable preCtxState is not smaller than 1, or after the temporary variable preCtxState is set to 1, the context initialization unit 112 determines whether the temporary variable preCtxState is larger than 126 (step S137).

  If the temporary variable preCtxState is larger than 126, the context initialization unit 112 sets 126 to the temporary variable preCtxState (step S138).

  If the temporary variable preCtxState is not greater than 126, or after 126 is set in the temporary variable preCtxState, the context initialization unit 112 determines whether the temporary variable preCtxState is 63 or less (step S139).

  When the temporary variable preCtxState is not 63 or less, the context initialization unit 112 is a value obtained by subtracting 64 from the temporary variable preCtxState to the state information pStateIdx that is a component of the context variable information corresponding to the currently initialized ctxIdx In addition, the context initialization unit 112 sets 1 to the dominant symbol information valMPS, which is a component of the context variable information (step S140).

If the temporary variable preCtxState is 63 or less, the context initialization unit 112 subtracts the temporary variable preCtxState from 63 to the state information pStateIdx that is a component of the context variable information corresponding to the currently initialized ctxIdx. Also, 0 is set to the dominant symbol information valMPS which is a component of the context variable information (step S141). Here, since preCtxState = 43,
[Equation 3]
pStateIdx = 63−preCtxState
= 63−43
= 20
valMPS = 0
It becomes.

  Next, the context initialization unit 112 stores the context variable information obtained in step S140 or step S141 in the context variable information storage unit 114 (step S142). Here, pStateIdx = 63−43 = 20 and valMPS = 0 are stored in the context variable information storage unit 114 as context variable information of ctxIdx = 70.

  Next, the context initialization unit 112 determines whether there is a ctxIdx to be initialized (step S143). If there is a ctxIdx to be initialized, the process returns to step S133 to continue the initialization process. If there is no ctxIdx to be converted, the context variable information initialization process is terminated.

  CABAC performs the initialization process of the context variable information through the above steps. Therefore, context initialization processing in CABAC depends on decoded syntax information such as pic_init_qp_minus26, slice_qp_delta, slice_type, and cabac_init_idc. Also, it is necessary to initialize a plurality of context variable information each time at the head of a slice, which is a CABAC decoding unit, based on the syntax information.

  Other JBIG and JPEG2000 also use a plurality of context variable information to improve the encoding efficiency.

  In JBIG, there are 113 context variable information similar to CABAC, and these context variable information is initialized at the beginning of decoding of an image as a decoding unit.

  Patent Document 1 is disclosed as a method for improving the initialization process in JBIG. In this Patent Document 1, when a plurality of pages of images are arithmetically encoded by a facsimile or the like, if the attribute of each image does not change, the state management variable at the end of arithmetic encoding of an image of a certain page, that is, state information, is dominant. When the symbol information is inherited and the attribute changes, the decoding side is notified that the state management variable is to be initialized, thereby reducing the initialization process and improving the compression efficiency.

  In the method of trying to reduce the initialization process and improve the compression efficiency disclosed in Patent Document 1, when the change of the attribute of each image is measured during encoding and the attribute of the image does not change, the context variable information If the state is inherited without initialization, and the attribute changes, the initialization process on the decoding side is reduced by notifying the decoding side that the context variable information is initialized. However, reducing the initialization process only on the decoding side is not considered, and depends on image attribute determination on the decoding side.

In addition, in the initialization process such as JBIG, it can be initialized with a predetermined constant in advance, but each context variable information necessary for arithmetic decoding such as the CABAC initialization process used in Non-Patent Document 4 is used. In other words, the state information and the dominant symbol information for specifying the state of the probability distribution are identified by the corresponding context initialization information by the syntax information that changes for each fixed decoding unit, and each context variable information is associated. While calculating based on each context initialization information, it had to be initialized every time every certain decoding unit.
Tomohiko Uematsu “Introduction to Document Data Compression Algorithm” CQ Publishing 1994 ITU-T TT. 82 ISO / IEC 15444 ITU-TH. H.264 JP-A-6-319046

  In order to solve the above problems, in the present invention, when context variable information is initialized when performing arithmetic decoding, the context variable information is accumulated as initialized context variable information, and a certain decoding unit is stored. When the context variable information that has been initialized by the acquired syntax information is accumulated as the initialized context variable information while monitoring the syntax information that changes every time, the initialized context variable information is stored in the initialized context variable information. Based on the context variable information is initialized. In this way, by reducing the number of operations when initializing context variable information in arithmetic decoding, the context variable information initialization process is reduced, the amount of calculation required for the initialization process is reduced, and the initialization process is performed. An object of the present invention is to provide a faster arithmetic decoding device and an arithmetic decoding program by reducing the time required for decoding to be completed by reducing the amount of calculation.

  In order to achieve the above object, the arithmetic decoding apparatus according to claim 1, wherein the arithmetic decoding apparatus searches for the input bitstream based on a start code for specifying a decoding position of the input bitstream, and performs arithmetic operation. The bit string to be decoded is specified, the specified bit string is stored in the buffer memory, and the header information of the bit string stored in the buffer memory is decoded based on the decoding position information indicating the position on the bit string at which the arithmetic decoding is started. Acquiring syntax information indicating a decoding procedure for arithmetically decoding a bit string at a specific position on the bit string specified by the decoding position information, and acquiring one of a plurality of context variable information prepared in advance, Initialization is performed according to the syntax information, and the initialized context variable information is used for the specific position. An arithmetic decoding device that arithmetically decodes a bit string, wherein a plurality of the context variable information initialized according to the type of syntax information is stored as initialized context variable information. And the decoded syntax information necessary for initializing the context variable information is acquired, and the context variable information is stored in the initialized context variable information storage unit according to the type of the decoded syntax information. If it is not stored, the context variable information is initialized according to the decoded syntax information, and then stored in the initialized context variable information storage means. , If stored, based on the decoded syntax information, the Characterized in that the initialized already a context variable information storage means and a context variable management means for acquiring specific the context variable information.

  The arithmetic decoding program according to claim 2 specifies a bit string to be arithmetically decoded by searching the input bitstream based on a start code for specifying a decoding position of the input bitstream. And decoding the header information of the bit string stored in the buffer memory based on the decoding position information indicating the position on the bit string where the specified bit string is stored in the buffer memory and starting the arithmetic decoding, and the decoding position information To obtain syntax information indicating a decoding procedure for arithmetic decoding of a bit string at a specific position on a bit string specified by the one of a plurality of context variable information prepared in advance according to the syntax information. And using the initialized context variable information, a bit string at the specific position An arithmetic decoding program for causing a computer to perform an arithmetic decoding device for arithmetic decoding, wherein the context variable information initialized according to the type of syntax information is initialized by the computer. A plurality of initialized context variable information storage means for acquiring the decoded syntax information necessary for initialization of the context variable information, and according to the type of the decoded syntax information, the context variable information Is stored in the initialized context variable information storage means, and if not stored, after initialization of the context variable information according to the decoded syntax information, Stored in the initialized context variable information storage means; If you were, on the basis of the decoded syntax information, characterized in that to function as the context variables management means for acquiring a particular said context variable information from the initialized context variable information storage means.

  In the present invention, by reducing the number of operations when initializing context variable information in arithmetic decoding, the initialization processing of context variable information can be reduced, and the amount of calculation required for the initialization processing can be reduced. Further, the time required for decoding to be completed can be shortened by reducing the amount of calculation required for the initialization process.

  An embodiment of the present invention will be described with reference to FIGS.

  FIG. 1 is a functional block diagram showing how an arithmetic decoding unit including an initialization unit of the present invention actually decodes a bitstream and outputs syntax information in an AVC decoder which is an arithmetic decoding device of the present invention. FIG.

  As shown in FIG. 1, the decoder 1 includes a bitstream acquisition unit 2, a decoding position detection unit 3, a bitstream buffer 4, a decoding position management unit 5, a syntax information acquisition unit 6, a syntax information storage unit 7, and an arithmetic The decoding unit 8 is used.

  The bit stream acquisition unit 2 has a function of acquiring an encoded bit stream for decoding.

  The decoding position detection unit 3 searches the bit stream acquired by the bit stream acquisition unit 2 based on the start code obtained from the decoding position management unit 5, and detects a position in the bit stream that matches the start code. Thus, it has a function of specifying the start position of decoding. The decoding position detection unit 3 has a function of specifying a decoding unit by specifying the next decoding position. Here, a bit stream that can decode at least one slice can be specified as a decoding unit, so that a better configuration can be obtained. The bit string in the bit stream specified in this way is stored in a bit stream buffer described later.

  The bit stream buffer 4 stores the bit string specified by the decoding position detection unit 3.

  The decoding position management unit 5 has a function of managing the decoding position in the input bitstream acquired by the bitstream acquisition unit 2. The decoding position management unit 5 has a function of managing the decoding position in the bit string stored in the bit stream buffer 4, and designates a start code to the decoding position detection unit 3 as necessary to specify a bit. The bit string in the stream buffer 4 is updated. Further, the decoding position management unit 5 notifies the syntax information acquisition unit 6 and the arithmetic decoding unit 8 of the decoding position information in the bit stream buffer 4 and detects the decoding position from the arithmetic decoding unit 8. In response to the request, the decoding position detector 3 has a function of notifying a start code corresponding to the detection request.

  The syntax information acquisition unit 6 is positioned higher than the bit sequence encoded by the arithmetic decoding unit 8 to be described later from the decoding location information obtained from the decoding location management unit 5 and the bit sequence stored in the bit stream buffer 4. The header information to be decoded is decoded, and the obtained syntax information is notified to the syntax information storage unit 7 described later.

  The syntax information storage unit 7 stores various types of syntax information obtained from the syntax information acquisition unit 6. In addition, the syntax information storage unit 7 has a function of notifying the arithmetic decoding unit 8 of syntax information as necessary.

  The arithmetic decoding unit 8 performs arithmetic decoding from the decoding position information obtained from the decoding position management unit 5 and the bit string stored in the bit stream buffer 4 from the position of the bit string specified by the decoding position information, and the decoding result is obtained. It has a function of outputting as output syntax information. Further, the arithmetic decoding unit 8 has a function of requesting the decoding position management unit 5 to detect a decoding position before or after the current decoding position as necessary.

  FIG. 2 is a block diagram showing the arithmetic decoding unit 8 of FIG. 1 in more detail.

  The arithmetic decoding unit 8 includes a state transition management unit 11, a context variable management unit 12, an initialized context variable information storage unit 13, a context initialization unit 14, a context initialization information storage unit 15, a context variable information storage unit 16, and a bit string. The count unit 17, the bit string information management unit 18, the context identification unit 19, the binary arithmetic decoding unit 20, the decoded bit acquisition unit 21, and the bit string comparison unit 22 are configured.

  The state transition management unit 11 has a function of managing the internal state in the arithmetic decoding unit 8 so that syntax information that is currently being decoded can be correctly decoded. The state transition management unit 11 performs an initialization request to the context variable management unit 12, control of the bit string count unit 17, control of the bit string information management unit 18, control of the context specification unit 19, and decoding to the binary arithmetic decoding unit 20. The internal state is managed by making a start request.

  The context variable management unit 12 has a function of determining whether initialization of context variable information is necessary in response to an initialization request from the state transition management unit 11. In initialization determination of context variable information, information initialized based on acquired decoded syntax information from initialized context variable information stored in an initialized context variable information storage unit 13 to be described later Search for the existence of. If the corresponding initialized context variable information exists as a result of the search, it is stored in the context variable information storage unit 16 to be described later, thereby reducing the number of computations when initializing the context variable information. If there is no corresponding initialized context variable information, an initialization request is made to the context initialization unit 14 in order to perform normal context variable information initialization processing, and the initialized context variable The information is stored as initialized context variable information in an initialized context variable information storage unit 13 described later and also stored in a context variable information storage unit 16 described later. Here, when storing context variable information in an initialized context variable information storage unit 13 to be described later, information for a predetermined number of prefetches is acquired from the decoded syntax information, and an initialized context variable information storage unit is acquired. It is possible to control the storage of context variable information with respect to 13, so that the configuration becomes better. When the decoded syntax information cannot be acquired by the predetermined number of prefetches, the prefetch request information 604 can be notified to the decoding position management unit 5 in FIG. 9, and the acquisition of the decoded syntax information can be attempted again.

  Further, when storing context variable information as initialized context variable information, the reference rate of the initialized context variable information stored in the initialized context variable information storage unit 13 is retained and newly initialized. When it becomes impossible to add to the context variable information storage unit 13, it can be replaced with the one with the lowest reference rate, so that a better configuration is obtained.

  The initialized context variable information storage unit 13 is initialized by the context initialization unit 14 and the context variable information notified by the context variable management unit 12 is based on the decoded syntax information acquired by the context variable management unit 12. And store it so that it can be reused without initializing the context variable information.

  The context initialization unit 14 has a function of performing initialization processing of context variable information used when performing arithmetic decoding in response to an initialization request from the context variable management unit 12. The context initialization unit 14 initializes context variable information using a predetermined constant stored in the context initialization information storage unit 15 and decoded syntax information notified from the context variable management unit 12. .

  Decoded syntax information necessary for CABAC initialization is pic_init_qp_minus26 included in pic_parameter_set_rbsp () and SliceQPY obtained from slice_qp_delta included in slice_header (), and slice_type and cabac_init_idc included in slice_header (). In CABAC, since the decoding unit is a slice, the context variable information is initialized at the head every time the slice is decoded.

  The context initialization information storage unit 15 stores a predetermined constant used when the context initialization unit 14 initializes context variable information. FIG. 12 is an excerpt of a part of a table in which predetermined constants used in CABAC are described as an example of the context initialization information stored in the context initialization information storage unit 15. In CABAC, constants m and n are identified from a table of context initialization information as shown in FIG. 12, using slice_type obtained from decrypted syntax information, cabac_init_idc, and ctxIdx for identifying each context. Used to initialize context variable information.

  The context variable information storage unit 16 stores context variable information that is actually used in arithmetic decoding. When the context variable information needs to be initialized, the context variable management unit 16 responds to a request from the state transition management unit 11. 12 has a function of acquiring necessary context variable information and notifying the context variable information storage unit 16. The context variable information storage unit 16 has a function of initializing context variable information by updating the context variable information stored based on the notified context variable information. The context variable information stored in the context variable information storage unit 16 is changed according to the decoding result of the binary arithmetic decoding unit 20.

  In the configuration of the arithmetic decoding unit 8 shown in FIG. 2, the context variable information is updated via the context specifying unit 19. In general, the binary arithmetic decoding unit 20 directly changes the contents of the context variable information storage unit 16. It is also possible to take a configuration that can be used.

  The bit string count unit 17 has a function of counting to which position in the bit string the bit to be decoded currently corresponds. This is because the occurrence probability of a bit differs depending on the bit position to be decoded, and therefore context variable information corresponding to the bit position needs to be associated.

  The bit string information management unit 18 has a function of converting information corresponding to syntax information candidates that are currently being decoded into information expressed in bit strings and performing association. This correspondence is used by the bit string comparison unit 22 described later.

  The context specifying unit 19 obtains the ctxIdx for specifying the context variable information of the bit currently being decoded from the information regarding the type of syntax information to be decoded currently obtained from the state transition management unit 11, from the bit string count unit 17. It has a function of specifying the bit position in the obtained bit string to be decoded and the decoded syntax information.

  The binary arithmetic decoding unit 20 has a function of performing arithmetic decoding using the context variable information specified by the context specifying unit 19. The context variable information whose state has been changed by arithmetic decoding is sequentially reflected in the context variable information storage unit 16. In the arithmetic decoding unit 8 in FIG. 2, this is performed via the context specifying unit 19. In addition, when bit supplementation from the bitstream is required by normalization at the time of arithmetic decoding, the binary arithmetic decoding unit 20 supplements necessary bits from the decoded bit acquisition unit 21.

  The decoded bit acquisition unit 21 has a function of acquiring a bit stream and decoding position information in response to a bit request from the binary arithmetic decoding unit 20.

  The bit string comparison unit 22 has a function of holding the output result of the binary arithmetic decoding unit 20 and comparing with the bit string information prepared by the bit string information management unit 18. If there is an equal bit string information, the bit string comparison unit 22 acquires syntax information corresponding to the equal bit string information, outputs the syntax information as output syntax information, and holds the binary arithmetic decoding unit 20 held therein. Is prepared for decoding of the next syntax information. In addition, the bit string comparison unit 22 notifies the state transition management unit 11 of output syntax information as necessary.

  FIG. 3 is a flowchart showing a processing procedure at the time of bitstream decoding. The outline of the operation of the decoder 1 during the decoding process will be described below with reference to FIG.

  First, when the decoder 1 starts the decoding process of the bit stream, the bit stream acquisition unit 2 acquires the input bit stream to be decoded in order to acquire the input bit stream (step S01).

  Next, the decoding position detection unit 3 acquires a start code necessary for specifying the decoding position from the decoding position management unit 5 (step S02). The decoding position detection unit 3 searches the bit stream acquired by the bit stream acquisition unit 2, identifies the bit string, and stores the identified bit string in the bit stream buffer 4 (step S03).

  Next, the syntax information acquisition unit 6 performs decoding based on the bit string stored in the bit stream buffer 4 and the decoding position information obtained from the decoding position management unit 5 in order to acquire syntax information. The syntax information is acquired (step S04), and the acquired syntax information is stored in the syntax information storage unit 7 (step S05).

  Based on the decoding position detection request notified from the arithmetic decoding unit 8 as necessary, the decoding position management unit 5 prefetches the bitstream and acquires the next syntax information when performing arithmetic decoding. It is determined whether it is necessary (step S06). When the next syntax information is necessary, the process proceeds to step S02.

  In order to perform arithmetic decoding, the arithmetic decoding unit 8 acquires a bit string from the bit stream buffer 4, acquires decoding position information of the bit string from the decoding position management unit 5, and obtains a position on the bit string specified from the decoding position information. In order to determine how to interpret and decode the bits, the necessary syntax information is obtained from the syntax information storage unit. The arithmetic decoding unit 8 performs arithmetic decoding based on these pieces of information, and sequentially outputs output syntax information that is a decoding result (step S07).

  Next, context variable information initialization processing will be described. FIG. 4 is a flowchart showing a processing procedure in the initialization processing of context variable information. Hereinafter, the operation of the arithmetic decoding unit 8 will be described with reference to FIG.

  When the initialization of the context variable information is started, the context variable management unit 12 acquires parameters necessary for the initialization of the context variable information (step S11). In FIG. 2, decoded syntax information acquired when the context variable management unit 12 receives an initialization request is a parameter necessary for initialization of context variable information.

  Next, the context variable management unit 12 determines whether there is context variable information initialized by the acquired parameter (step S12). In the arithmetic decoding unit 8 shown in FIG. 2, the context variable management unit 12 performs initialization stored in the initialized context variable information storage unit 13 based on the decoded syntax information that is the parameter acquired in step S11. Search for completed context variable information. When the context variable information initialized with the acquired parameter is stored in the initialized context variable information storage unit 13, the process proceeds to step S15, and otherwise, the process proceeds to step S13.

  When the context variable information initialized with the acquired parameter is not stored in the initialized context variable information storage unit 13, the context variable management unit 12 issues an initialization request to the context initialization unit 14 (step S13). In response to this initialization request, the context initialization unit 14 performs normal context variable information initialization processing.

  The context variable management unit 12 acquires the context variable information initialized by the context initialization unit 14 and stores it in the initialized context variable information storage unit 13 (step S14).

  Next, the context variable management unit 12 searches the initialized context variable information stored in the initialized context variable information storage unit 13 on the basis of the acquired decoded syntax information, thereby corresponding context variables. Information is specified (step S15).

  Next, the context variable management unit 12 stores the context variable information specified in step S15 in the context variable information storage unit 16 (step S16).

  In this way, initialization processing of context variable information necessary at the time of decoding is performed.

  Next, context variable information initialization processing will be described in detail based on the flowchart of FIG.

  When the initialization of the context variable information is started, the context variable management unit 12 responds to the initialization request from the state transition management unit 11, and the pic_init_qp_minus26, slice_qp_delta, slice_type, slice_type, cabac_init_idc is acquired (step S21). Here, it is assumed that pic_init_qp_minus26 = 4, slice_qp_delta = 5, slice_type = 0 (P slice), and cabac_init_idc = 1 are acquired.

  Here, when a predetermined number of prefetches for controlling the storage of context variable information in the initialized context variable information storage unit 13 is set, the context variable management unit 12 determines a predetermined number from the decoded syntax information. Get information for the number of read-aheads. When the decoded syntax information cannot be acquired by a predetermined number of prefetches, the context variable management unit 12 notifies the prefetch request information to the decoding position management unit 5 and acquires the decoded syntax information again.

  Next, in order to search for the initialized context variable information corresponding to the syntax information that is currently being decoded, the context variable management unit 12 determines the initialized context variable based on the type of the acquired decoded syntax information. It is searched whether there is context variable information corresponding to the decoded syntax information acquired from the initialized context variable information stored in the information storage unit 13 (step S22), and whether there is initialized context variable information. Is determined (step S23). If the corresponding initialized context variable information exists, the process proceeds to step S24. Otherwise, the process proceeds to step S25.

  When the corresponding initialized context variable information exists, the context variable management unit 12 stores the context variable information specified from the initialized context variable information storage unit 13 in the context variable information storage unit 16 (step S24), End the initialization process of the context variable information.

  When there is no corresponding initialized context variable information, the context variable management unit 12 calculates SliceQPY according to the following mathematical formula (step S25).

[Equation 4]
SliceQPY = 26 + pic_init_qp_minus26 + slice_qp_delta
= 26 + 4 + 5
= 35
Next, the context initialization unit 14 identifies ctxIdx that needs to be initialized from the decoded context variable information slice_type and cabac_init_idc acquired in step S21, and sets m and n corresponding to the ctxIdx to the context initialization information storage unit. 15 (step S26). Here, assuming that ctxIdx requiring initialization is ctxIdx = 70, m = 13 and n = 15 are acquired according to the context initialization information table shown in FIG.

  Next, the context initialization unit 14 calculates a temporary variable preCtxState according to the following equation using SliceQPY calculated in step S25 and m and n acquired in step S26 (step S27).

[Equation 5]
preCtxState = ((m × SliceQPY) >> 4) + n
= (13 × 35) >> 4 + 15
= 28 + 15
= 43
Here, “>> 4” in the equation indicates that a 4-bit right shift operation is performed.

  Next, the context initialization unit 14 determines whether or not the temporary variable preCtxState is smaller than 1 (step S28). If the temporary variable preCtxState is smaller than 1, the process proceeds to step S29. Otherwise, the process proceeds to step S30.

  When the temporary variable preCtxState is smaller than 1, the context initialization unit 14 sets 1 to the temporary variable preCtxState (step S29).

  Next, the context initialization unit 14 determines whether or not the temporary variable preCtxState is greater than 126 (step S30). If the temporary variable preCtxState is greater than 126, the process proceeds to step S31. Otherwise, the process proceeds to step S32.

  When the temporary variable preCtxState is larger than 126, the context initialization unit 14 sets 126 to the temporary variable preCtxState (step S31).

  Next, the context initialization unit 14 determines whether the temporary variable preCtxState is 63 or less (step S32). If the temporary variable preCtxState is 63 or less, the process proceeds to step S34. Otherwise, the process proceeds to step S33.

  If the temporary variable preCtxState is not 63 or less, the context initialization unit 14 sets a value obtained by subtracting 64 from the temporary variable preCtxState in the state information pStateIdx that is a component of the context variable information corresponding to the currently initialized ctxIdx. In addition, 1 is set to the dominant symbol information valMPS which is a component of the context variable information (step S33).

  When the temporary variable preCtxState is 63 or less, the context initialization unit 14 subtracts the temporary variable preCtxState from 63 to the state information pStateIdx that is a component of the context variable information corresponding to the currently initialized ctxIdx. Is set, and the dominant symbol information valMPS, which is a component of the context variable information, is set to 0 (step S34).

Here, since preCtxState = 43,
[Equation 6]
pStateIdx = 63−preCtxState
= 63−43
= 20
valMPS = 0
It becomes.

  Next, the context initialization unit 14 stores the context variable information acquired in step S33 or step S34 in the context variable information storage unit 16 (step S35). Further, the context initialization unit 14 stores similar context variable information in the initialized context variable information storage unit 13 as initialized context variable information.

  Here, pStateIdx = 63−43 = 20 and valMPS = 0 are stored in the context variable information storage unit 16 and the initialized context variable information storage unit 13 as context variable information of ctxIdx = 70.

  Here, when a predetermined amount of prefetching is set, and in step S21, when the decoded syntax information can be acquired by the amount of the predetermined prefetching amount, the condition for initializing the current context variable information is: By storing only when it corresponds to the acquired decoded syntax information, and discarding it without storing it, the storage to the initialized context variable information storage unit 13 is controlled.

  The context initialization unit 14 determines whether there is a ctxIdx to be initialized (step S36). If there is a ctxIdx to be initialized, the process proceeds to the process of step S26, and the initialization process is continued. If there is no ctxIdx to be initialized, the context variable information initialization process is terminated.

  In this way, the initialization processing of unnecessary context variable information is reduced, the amount of computation required for the initialization processing is reduced, and the time required for decoding to be completed is reduced by reducing the amount of computation required for the initialization processing. Can be shortened. In particular, by including the context variable management unit 12 and the initialized context variable information storage unit 13, it is possible to reuse the context variable information that has been initialized each time for a certain decoding unit.

  Next, CABAC decoding processing in the present embodiment will be described based on the flowchart of FIG. FIG. 6 is a flowchart showing a decoding operation for one slice, which is a CABAC decoding unit.

  When the CABAC decoding process is started, the state transition management unit 11 determines whether the syntax information to be decoded is the first syntax information in the slice (step S41). If the syntax information to be decoded is the first syntax information, the process proceeds to step S42. If the syntax information is not the first syntax information, the process proceeds to step S44.

  When the syntax information to be decoded is the first syntax information, the state transition management unit 11 makes an initialization request to the context variable management unit 12, and the context initialization unit 14 When an initialization request is issued from 11, the above-described context variable information is initialized (step S42).

  When the state transition management unit 11 issues an initialization request to the binary arithmetic decoding unit 20, the binary arithmetic decoding unit 20 initializes the range to a predetermined value and sets the value from the decoded bit acquisition unit 21. It initializes with the acquired bit (step S43).

  Next, when the state transition management unit 11 requests the bit string information management unit 18 to convert information associated with syntax information candidates to be currently decoded into information expressed in bit strings, the request is based on this request. The bit string information management unit 18 creates bit string information (step S44).

  Next, the state transition management unit 11 initializes the bit string counter binIdx in the bit string counting unit 17 to −1 (step S45).

  Next, the state transition management unit 11 increments the bit string counter binIdx in the bit string counting unit 17 by 1 (step S46).

  Next, when the state transition management unit 11 requests the context determination unit 19 to specify the ctxIdx for specifying the context variable information used for decoding, the context determination unit 19 uses the bit currently being decoded. CtxIdx for specifying the context variable information of information on the type of syntax information to be decoded currently obtained from the state transition management unit 11, the bit position in the bit string to be decoded currently obtained from the bit string counting unit 17 Then, identification is performed using the decoded syntax information (step S47).

  Next, when the state transition management unit 11 requests the binary arithmetic decoding unit 20 to decode, the binary arithmetic decoding unit 20 obtains context variable information obtained based on the ctxIdx specified from the context specifying unit 19. Obtain and decrypt (step S48). When bit supplementation from the bitstream is required by normalization at the time of arithmetic decoding, the binary arithmetic decoding unit 20 supplements necessary bits from the decoded bit acquisition unit 21, and when the decoding is completed, the decoding result To the bit string comparison unit 22.

  The bit string comparison unit 22 compares the held output result from the binary arithmetic decoding unit 20 with the bit string information prepared by the bit string information management unit 18, and there is an equal bit string information. Whether or not (step S49). If there is no equal bit string information, the process proceeds to step S46 and the process is continued. If there is an equal bit string information, the bit string comparison unit 22 acquires syntax information corresponding to the equal bit string information, outputs the syntax information as output syntax information, and holds the binary arithmetic decoding unit 20 held therein. In response to the initialization of the output result, the state transition management unit 11 is notified of the output syntax information in preparation for decoding of the next syntax information.

  Next, the state transition management unit 11 determines whether or not the next syntax information to be decoded that is currently the decoding target is mb_type and the decoding result obtained in the process of step S48 is I_PCM. To do. If this condition is satisfied, the process proceeds to step S51. If the condition is not satisfied, the CABAC decoding is terminated.

  In step S51, the state of the binary arithmetic decoding unit 20 is initialized by performing the same process as in step S43. When the initialization is completed, the CABAC decryption is terminated.

  By performing the processing in this way, the number of operations when initializing the context variable information is reduced, and it is possible to speed up the initialization processing at the time of arithmetic decoding.

  The present invention includes a program for causing a computer to realize the functions of the arithmetic decoding device (decoder 1). This program may be read from a recording medium and taken into a computer, or may be transmitted via a communication network and taken into a computer.

It is a functional block diagram which shows the basic composition of the decoder of this invention. It is a functional block diagram which shows the structure of the arithmetic decoding part of this invention. It is a flowchart which shows the basic operation | movement of this invention. It is a flowchart which shows the basic operation | movement of the arithmetic decoding part of this invention. It is a flowchart which shows the detailed operation | movement of the arithmetic decoding part of this invention. It is a flowchart which shows operation | movement of the arithmetic decoding part of this invention. It is a conceptual diagram for demonstrating the concept of arithmetic encoding / decoding. It is a functional block diagram which shows the basic composition of CABAC. It is a functional block diagram which shows the structure of the arithmetic decoding part in CABAC. It is a figure for demonstrating the isolation | separation of the series by context switching. It is a flowchart which shows the basic operation | movement of CABAC. It is a figure for showing the example of parameters m and n used when initializing context variable information in CABAC. It is a flowchart which shows the operation | movement of the arithmetic decoding part in CABAC. It is a flowchart which shows the operation | movement in the initialization process of the context variable information in CABAC.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Decoder 2 Bit stream acquisition part 3 Decoding position detection part 4 Bit stream buffer 5 Decoding position management part 6 Syntax information acquisition part 7 Syntax information storage part 8 Arithmetic decoding part 11 State transition management part 12 Context variable management part 13 Initialization Context variable information storage unit 14 Context initialization unit 15 Context initialization information storage unit 16 Context variable information storage unit 17 Bit string count unit 18 Bit string information management unit 19 Context identification unit 20 Binary arithmetic decoding unit 21 Decoding bit acquisition unit 22 Bit string comparison unit 101 Decoder 102 Bit stream acquisition unit 103 Decoding position detection unit 104 Bit stream buffer 105 Decoding position management unit 106 Syntax information acquisition unit 107 Syntax information storage unit 108 Arithmetic decoding unit 111 State Transition Management Unit 112 Context Initialization Unit 113 Context Initialization Information Storage Unit 114 Context Variable Information Storage Unit 115 Bit Sequence Count Unit 116 Bit Sequence Information Management Unit 117 Context Specification Unit 118 Binary Arithmetic Decoding Unit 119 Decoded Bit Acquisition Unit 120 Bit Sequence Comparison part

Claims (2)

Based on the start code for specifying the decoding position of the input bit stream, the input bit stream is searched to specify a bit string for arithmetic decoding, and the specified bit string is stored in the buffer memory,
Based on decoding position information indicating a position on the bit string at which arithmetic decoding is started, the header information of the bit string stored in the buffer memory is decoded, and a bit string at a specific position on the bit string specified by the decoding position information To obtain syntax information indicating a decoding procedure for arithmetic decoding,
Arithmetic decoding that initializes one of a plurality of context variable information prepared in advance according to the syntax information, and arithmetically decodes the bit string at the specific position using the initialized context variable information Device.
Initialized context variable information storage means for storing a plurality of the context variable information initialized according to the type of syntax information as initialized context variable information;
The decoded syntax information necessary for initialization of the context variable information is acquired, and the context variable information is stored in the initialized context variable information storage unit according to the type of the decoded syntax information. Whether or not
If not stored, after initializing the context variable information according to the decoded syntax information, stored in the initialized context variable information storage means,
If stored, context variable management means for acquiring the specific context variable information from the initialized context variable information storage means based on the decoded syntax information;
An arithmetic decoding device comprising: Based on the start code for specifying the decoding position of the input bit stream, the input bit stream is searched to specify a bit string for arithmetic decoding, and the specified bit string is stored in the buffer memory,
Based on decoding position information indicating a position on the bit string at which arithmetic decoding is started, the header information of the bit string stored in the buffer memory is decoded, and a bit string at a specific position on the bit string specified by the decoding position information To obtain syntax information indicating a decoding procedure for arithmetic decoding,
Arithmetic decoding that initializes one of a plurality of context variable information prepared in advance according to the syntax information, and arithmetically decodes the bit string at the specific position using the initialized context variable information An arithmetic decoding program for causing a computer to realize the encoding device,
The computer,
Initialized context variable information storage means for storing a plurality of the context variable information initialized according to the type of syntax information as initialized context variable information;
The decoded syntax information necessary for initialization of the context variable information is acquired, and the context variable information is stored in the initialized context variable information storage unit according to the type of the decoded syntax information. Whether or not
If not stored, after initializing the context variable information according to the decoded syntax information, stored in the initialized context variable information storage means,
If stored, context variable management means for acquiring the specific context variable information from the initialized context variable information storage means based on the decoded syntax information;
An arithmetic decoding program characterized by being made to function.

Images