JPH04274674A - Data converter - Google Patents

Abstract

PURPOSE:To split a code data an optional hierarchical number without increasing the hardware scale by implementing plural number of times of hierarchical split when a sequential coding data is converted into a hierarchical coding data. CONSTITUTION:Number of coefficients of each layer in a block comprising NXN picture elements is set by data quantity setting sections 106, 107. Number of zero coefficients and non-null coefficients are decoded by a decoding table 101 from an input sequential code data to detect consecutive zero coefficients between layers by a detection section 111. Number of zero coefficients belonging to each hierarchy is calculated from consecutive zero coefficients existing between hierarchys. Number of zero coefficients in response to the consecutive/non-consecutive zero coefficients between layers and non-null coefficients are coded by a demultiplexer section 115. A variable length code data generated continuously is subjected to packing by packing sections 121, 122 to covert the data into M sets of hierarchical codes. In the conversion into the hierarchical data in excess of M hierarchys, the variable length code data is converted into a hierarchical code data whose hierarchy is M or less and the code data of a hierarchy not split is converted, and the processing is repeated to obtain the split hierarchical code.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a code data converting apparatus in a variable length coding system for coding the number of consecutive zeros and non-zeros.

0002

2. Description of the Related Art In recent years, means for encoding and compressing data has become an important issue in signal processing of still images.

Conventionally, as a still image encoding method, focusing on image correlation, image data is divided into blocks each consisting of N×N pixels, and the data in the blocks is subjected to discrete cosine transform (DCT), etc. There is a method that performs an orthogonal transform and encodes and compresses the transform coefficients. FIG. 15 is a schematic diagram showing the arrangement of transform coefficients subjected to discrete cosine transform on N×N pixels in a block. The shaded area is the area where the numerical value is large. As shown in the figure, the transform coefficients of blocks with many low frequency components are concentrated in the upper left of the block. Therefore, these transform coefficients are arranged in one dimension, and the number of consecutive zero coefficients and non-zero coefficient values are encoded for the arranged transform coefficients. The encoding process will be explained below. For example, when encoding transform coefficients obtained by orthogonally transforming an 8 x 8 pixel block as shown in Figure 16, the transform coefficients are scanned along the route and direction indicated by the polygonal lines in the figure, arranged one-dimensionally, and then The number of zero coefficients and non-zero coefficient values are encoded. Furthermore, if zero coefficients are arranged continuously from after the pixel marked 1 in FIG. 17 to the final pixel, that part is not encoded and an EOB (End Of Block) code is added. FIG. 18 schematically shows codes corresponding to the coefficient array in such a case. In this way, the encoding process is performed on a block-by-block basis, and this block-by-block process is applied to the entire image. Hereinafter, the code in which this block is a processing unit will be referred to as sequential code data.

On the other hand, in a system that handles images, it is necessary to handle images hierarchically. For example, FIG. 4 schematically shows how a block is divided into three layers. For this purpose, it is necessary to manage code data hierarchically. In this case, it is necessary to convert from a sequential code to a layered code, but this conversion process is performed by decoding the sequential code data once to obtain coefficients, and then sequentially encoding the obtained coefficients for each layer. It will be done. Additionally, in order to decode a layered code into a sequential code, it is necessary to read and decode the layered code data up to the necessary layers, and then re-encode the coefficients up to the necessary layers and convert them into sequential code data. be.

FIG. 6 schematically shows the process of converting sequential code data into hierarchical code data by dividing it into three layers, and FIG. 14 schematically shows the storage state of the hierarchical code data. As mentioned above, sequential code data is encoded by sequentially handling the coefficient data A, B, and C in each block. On the other hand, in hierarchical code data, the coefficient data A, B, and C in each block are encoded. C is treated and encoded independently and managed hierarchically. The flow of the conversion process will be described with reference to FIG. 6. First, the sequential code of block BLK00 is decoded to obtain coefficients A, B, and C within the block. Next, the obtained coefficient data is sequentially encoded. In this case, the coefficient data of A is encoded to find the sign of BLK00-A, and then,
The coefficient data of B is encoded to obtain the code of BLK00-B, and then the coefficient data of C is encoded to obtain the code of BLK00-C, and each layer is packed and stored. On the other hand, FIG. 11 schematically shows the process of converting hierarchical code data into sequential code data up to the required layers, taking the case of A and B layers as an example. To explain BLK00, code data BLK00-A of layer A and code data BLK00-B of layer B are sequential code data BL.
Convert to K00-AB. In this case, the encoded data of layer A and the encoded data of layer B are unpacked separately,
It is decoded to obtain coefficient groups A and B. This coefficient group A and B are B
Continuously handled and encoded within LK00, encoded data B
Converted to LK00-AB. As shown in FIG. 12, this sequential code data BLK00-AB is code data obtained by coding the coefficient data of the C layer as all zero coefficients. Code data B layered in this way
LK00-A and BLK00-B are converted to BLK00-AB, which is obtained by sequentially encoding the coefficient data shown in FIG. 12.

[0006]

[Problems to be Solved by the Invention] In such a conventional encoded data conversion device, (1) when converting sequential encoded data into hierarchical encoded data, if the number of layer divisions increases, the code is changed for each layer. Since a data packing processing section is required, the hardware scale becomes large, and if the number of code data packing processing sections for each layer is limited, the number of layer divisions is limited.

(2) When converting hierarchical code data into sequential code data up to the required layer, if the number of layer divisions increases, an unpacking processing section for code data is required for each layer, so the hardware The size of the wear increases.

(3) When converting sequential code data to hierarchical code data, by dividing it into layers, runs with zero coefficients are divided, and the frequency of occurrence of runs with zero coefficients is different between sequential code and hierarchical code. The compression ratio decreases because the case is different.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a code data conversion device that can effectively execute code data conversion processing while reducing the hardware scale.

0010

[Means for Solving the Problems] In order to achieve the above object, the present invention provides a data amount setting means for setting the number of coefficients of each layer in a block consisting of N×N pixels as a first means of solving the problem. , a decoding means for obtaining the number of zero coefficients and a non-zero coefficient value from input code data, a detection means for detecting consecutive zero coefficients between layers, and a decoding means for detecting consecutive zero coefficients between layers belonging to each layer A calculation means for calculating the number of zero coefficients, and if zero coefficients are continuous between layers, encode the number of zero coefficients and non-zero coefficient values obtained by the calculation means, and if zero coefficients are continuous between layers, If not, an encoding means for encoding the number of zero coefficients and non-zero coefficient values obtained by the decoding means, and variable length code data continuously generated in the order of the layers in the encoding means, for each layer. It has packing means capable of converting into a hierarchical code with a number of layers of M (M: a natural number), and when converting to hierarchical data with a number of layers exceeding M, a packing means with a number of layers of less than M is provided. After converting to layered code data, the undivided layer code data is further converted to M
The data converting device repeats the process of converting into hierarchical code data with a number of layers equal to or less than 1, and obtains a hierarchical code divided into an arbitrary number of layers.

A second means for solving the problem is a data amount setting means for setting the number of coefficients belonging to each layer in a block consisting of N×N pixels, and packing for packing at least one or more variable length code data. means and
A code data storage means for temporarily storing code data in the middle of packing in each layer, and a code word of a layer currently being processed is processed by the packing means, and a code word of a layer whose processing is interrupted is stored in the code data storage means. The data converting apparatus may include a control means for temporarily storing the code word and, when processing is resumed, taking out the code word in the middle of processing from the code data storage means and controlling the code word to be processed by the packing means.

As a third means for solving the problem, a data amount setting means for setting the number of coefficients of each layer in a block consisting of N×N pixels, a layer number setting means for setting the number of layers to be reproduced, and a hierarchical type Extract the code data in hierarchical order,
address generation means for unpacking code data packed for each layer into variable-length code words and generating an address for the code word; code data for temporarily storing the code word being unpacked by the address generation means; The storage means and the code word of the layer currently being processed are processed by the address generation means, the code word of the layer whose processing is interrupted is stored in the code data storage means, and when the processing is resumed, the code word of the layer being processed is stored in the code data storage means. Extract the codeword that is being processed from
A control means for controlling processing by the address generation means, a decoding means for obtaining the number of zero coefficients and a non-zero coefficient value from the hierarchical code data, and a detection means for detecting consecutive zero coefficients between layers. , a calculation means for calculating the number of consecutive zero coefficients between layers from the zero coefficients belonging to each layer, and when zero coefficients are continuous between layers, the number of zero coefficients and non-zero coefficient values obtained by the calculation means. The data conversion apparatus is provided with an encoding means for encoding the number of zero coefficients and the non-zero coefficient value obtained by the decoding means when the zero coefficients are not consecutive between layers.

[0013]

[Operation] In the first means of solving the problem, the present invention, when converting sequential encoded data into hierarchical encoded data, divides it hierarchically in multiple times if it cannot be divided at once for the number of layers to be divided. By doing this, it can be divided into any number of layers.

In the second means for solving the problem, when sequential code data is converted into hierarchical code data, only the code words of the layer being processed are processed by the packing means, and the code words of the layer whose processing is interrupted are processed by the packing means. Temporarily stored in storage means,
When the processing is restarted, the code word that is being processed is retrieved from the storage means and processed by the packing means, thereby processing data conversion by an arbitrary number of hierarchical divisions at once.

[0015] In the third means for solving the problem, when converting hierarchical code data into sequential code data, only the code word of the layer that is being processed is processed by the address generation means, and the code word of the layer that is currently being processed is processed by the address generating means. is temporarily stored in the storage means, and when the processing is restarted, the code word being processed is retrieved from the storage means and processed by the address generation means, thereby processing data conversion by an arbitrary number of hierarchical divisions at once.

[0016]

Embodiments (Embodiment 1) Hereinafter, a data conversion device according to an embodiment of the first problem-solving means of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a code data converter for converting sequential code data into hierarchical code data in a data converter according to an embodiment of the first problem-solving means of the present invention. In the figure, 100 is an address generation unit that generates addresses from sequential code data, 101 is a decoding table, 102 is a register unit that temporarily stores continuous zero run lengths, 103 is a register unit that temporarily stores non-zero coefficient values, and 104 105 is a decoder unit that detects the number of coefficients from the zero run length of the register unit 102 and the non-zero coefficient value of the register unit 103; and 105 is an address generation unit 100.
106 and 107 are intra-layer data amount setting sections that set the number of coefficients in each layer in the block; 108 is intra-layer data amount setting sections 106 and 107
A subtraction unit 109 selects one of the data amounts set in and calculates the number of remaining data in the processing layer, and 109 calculates the number of unprocessed coefficients in the layer being processed when there are consecutive zero coefficients between layers. , a subtraction unit 110 that calculates the number of zero coefficients belonging to each layer from the zero run length stored in the register 102;
111 is a detection unit that detects consecutive zero coefficients between layers; 112 is a register unit that temporarily stores the output of the subtraction unit 109; and 113 is a layer control that controls each layer. part, 114 is a code table, 11
5 is a demultiplexer unit 116 that demultiplexes the output from the code table 114 according to the layer being processed;
and 117 are register units for temporarily storing code words of each layer;
118 and 119 are register units that temporarily store the code length;
20 is a code length control unit that controls the code length, 121 and 122
123 is a code data packing unit that packs the code data of each layer for each layer, and 123 is a multiplexer unit that selects and outputs either code data packing unit 121 or 122. Furthermore, the code table 114 may be a different code table from the code table for sequential codes.

The mutual relationships and operations of the above-mentioned components will be explained below using the case of a block consisting of 8×8 pixels as shown in FIG.
, FIG. 7, FIG. 8, and FIG. 10. Figure 5 and Figure 7 show that 3 of A, B1, and B2 are processed twice in the block.
A schematic diagram shows the process of dividing into two layers. First, in the first process, the inside of the block is divided into two layers, A and B, to obtain hierarchical code data of two layers. Furthermore, in the second process, B
The hierarchy is divided into two hierarchies, B1 and B2, and the whole hierarchy is divided into three hierarchies. FIG. 8 shows a state in which an 8×8 pixel block is divided into three layers. In the figure, the DC component shown in the upper left is data indicating the average value of this block, and since this DC component is encoded independently, it will not be described here. It is also assumed that K1 to K4 written at the pixel positions are non-zero coefficient values, and the coefficient values of other pixels are zero coefficients. In sequential encoding, the coefficient values are scanned and encoded in a zigzag manner from the low frequency component to the high frequency component as shown in FIG. 16, excluding the DC component. FIG. 10(a) shows sequential encoded data obtained by encoding the blocks shown in FIG.

In the block diagram of FIG. 1, FIG. 10(a)
sequential code data is input in sequence. first,
The sequential code data in FIG. 10(a) is A, (B1
The operation of dividing the data into two layers (+B2) and obtaining the hierarchical code data shown in FIG. 10(b) will be described. First, the amount of data in each layer is 5 for the A layer and 58 for the (B1+B2) layer, so the intra-layer data amount setting unit 1
06 and 107 are set to "5" and "58". next,
A sequential code is input to the address generation unit 100 of FIG. 1, and the address for S-CODE1 is given to the generated decoding table 101. Decoding table 101
is 3 zero runs and non-zero coefficient value “K” for S-CODE1.
1” is output, and the zero run length “3” is output to the register section 102.
The non-zero coefficient value “K1” is taken into the register section 103. The subtraction unit 108 selects the contents of layer A in the intra-layer data amount setting unit 106, and the detection unit 111 selects “5” in the intra-layer data amount setting unit 106 and the run length “3” in the register unit 102.
The subtraction unit 109, the register unit 110, and the register unit 112 are notified that the data is of layer A, and the data in the register unit 102 is taken into the register unit 112, and the data in the register unit 103 is taken into the register unit 110. In this case, the register section 112 takes in the zero run length "3", and the register section 110 takes in the non-zero coefficient value "K1". code table 114
takes in the data of the register section 112 and the register section 110 as an address, and generates the code data P-CODE-A-1.
and the code length. This data is demultiplexed by the demultiplexer section 115, and the code word register section 11
6 and is taken into the code length register section 118. The data in the code word register section 116 is stored in the packing section 12 under the control of the code length control section 120 as layer A data.
1, and when a predetermined amount of data is accumulated, it is output.

Next, address generation section 100 shifts the data by the code length of S-CODE1 under the control of address generation control section 105, generates an address for S-CODE2, and provides it to decoding table 101. The decoding table 101 outputs 13 zero runs and a non-zero coefficient value "K2" for S-CODE2, the zero run length "13" is taken into the register section 102, and the non-zero coefficient value "K2" is taken into the register section 103. , the layer control unit 11 via the decoder unit 104
It will be incorporated into 3. The hierarchy control unit 113 outputs the number of coefficients processed last time, “4”, and provides it to the subtraction unit 108. The subtraction unit 108 takes the difference between the content of the A layer of the intra-layer data amount setting unit 106, “5”, and the output “4” from the hierarchy control unit, and outputs “1”. The detection unit 111 compares the content “1” of the subtraction unit 108 with the content of the register unit 102, informs the subtraction unit 109, the register unit 110, and the register unit 112 that the A and B layers are continuous, and Part 1
10 and the register section 112 are set to "0" and given to the code table 114 as an address. code table 1
14 outputs the EOB code and code length. This data is demultiplexed by the demultiplexer section and taken into the code word register section 116 and code length register section 118. The data in the code word register section 116 is stored in the packing section 1 as A-layer data under the control of the code length control section 120.
21, and when a predetermined amount of data is accumulated, it is output. Next, the subtraction section 109 takes the difference between the zero run length "13" of the register section 102 and "1" of the subtraction section 108, outputs a difference value "12", and takes it into the register section 112. The code table 114 contains the zero run length of the register section 112.
12” and the non-zero coefficient value “K2” of the register section 110 are given as addresses, and the code data P-CODE-B-1
and the code length. This data is demultiplexed by the demultiplexer section 115, and the code word register section 11
7 and is taken into the code length register section 119. The data in the code word register section 117 is packed as B layer data by the packing section 122 under the control of the code length control section 120, and is output when a predetermined amount of data is accumulated. In this way, encoded data is obtained hierarchically. A layer, (
A hierarchical code divided into two layers (B1+B2) layer is shown in FIG. 10(b).

When redividing the hierarchical code data of the (B1+B2) layer obtained in the above manner into the B1 layer and the B2 layer, the data amount of the B1 layer is set to "30" in the intra-layer data amount setting units 106 and 107. This can be achieved by setting the data amount of the B2 layer to "28" and inputting the code of the (B1+B2) layer as a sequential code. The (B1+B2) hierarchy is subdivided into the B1 hierarchy and the B2 hierarchy, and the data converted into a hierarchical code is shown in FIG. 10-3.

Note that the code table can be changed depending on the number of layers to be divided and the number of coefficients in each layer, or the code table can be changed for each layer.

(Embodiment 2) A data conversion device according to an embodiment of the second problem-solving means of the present invention will be described below with reference to the drawings.

FIG. 2 is a block diagram showing the configuration of a code data converter for converting sequential code data into hierarchical code data in the data converter of the second problem solving means of the present invention. In the figure, 200 is an address generation unit that generates an address from sequential code data;
1 is a decoding table, 202 is a register section that temporarily stores continuous zero run lengths, 203 is a register section that temporarily stores non-zero coefficient values, and 204 is the zero run length of register section 202 and the non-zero coefficient value of register section 203. 205 is an address generation control unit that controls the address generation unit 200; 206 is an intra-layer data amount setting unit that sets the data amount of each layer that can be divided within the block; 208 209 is a subtraction unit that selects one of the data amounts set in the intra-layer data amount setting unit 206 and calculates the remaining data number of the processing layer; 209 is an intra-layer data amount setting unit 20
210 is a register unit that temporarily stores non-zero coefficient values; 210 is a register unit that temporarily stores non-zero coefficient values; 211 is a comparison unit that compares the data amount of each layer set by the intra-layer data amount setting unit 206 and the zero run length stored in the register 202; 212 is a subtraction unit 2;
213 is a layer control section that controls each layer; 214 is a code table; 215
216 is a register unit that temporarily stores the code words of each hierarchy; 218 is a register unit that temporarily stores the code length; 220 is a register unit that temporarily stores the code length; A code length control unit 221 is a code data packing unit that packs code data of each layer for each layer, and 222 is a memory unit that temporarily stores code data that is being packed.

The mutual relationship and operation of the above-mentioned components will be explained below with reference to FIGS. 8 and 10. Figure 10(a
) indicates sequential encoded data obtained by encoding the blocks shown in FIG. In FIG. 2, sequential code data shown in FIG. 10(a) is inputted one after another. First, the operation of dividing the sequential code data into three layers A, B1, and B2 to obtain the hierarchical code data shown in FIG. 10(c) will be described. The amount of data in each layer is “5” for layer A.
, the B1 layer is “30” and the B2 layer is “28”, so “5”, “30” and “28” are set in the intra-layer data amount setting unit 106.

Next, the sequential code is input to address generation section 200 shown in FIG. Thereafter, the same operation as in FIG. 1 is performed, and the code word of layer A is taken into the code register section 216.

During the processing of layer A, the code register section 216
The codes are sequentially fetched into the packing unit 221 from then on, and when a predetermined amount is accumulated, the codes are output. When the process moves to the B1 layer, the code data of the packing unit 221 is written to the memory 222, and the process continues. In this case, by preparing a plurality of code tables in the code table 214, the code table can be dynamically switched at each layer. The B1 layer code words taken into the code register section 216 are sequentially taken into the packing section 221, and when a predetermined amount is accumulated, they are output.

Next, when the process moves to the B2 layer, the code data of the packing section 221 is written into the memory 222, and the process continues. When the processing of one block is completed and the processing moves to the next block, the code data that is being processed in layer A is read out from the memory unit 222 to the packing unit 221, and the processing is continued.

Note that the code table can be changed for each layer depending on the number of layers to be divided and the number of coefficients in each layer.

(Third Embodiment) A data conversion device according to an embodiment of the third problem-solving means of the present invention will be described below with reference to the drawings.

FIG. 3 is a block diagram showing the configuration of a code data converter for converting hierarchical code data into sequential code data in a data converter according to an embodiment of the third problem-solving means of the present invention. In the figure, 300 is an address generation unit that generates an address from code data, 301 is an address control unit, 302 is a memory unit that temporarily stores code data during unpacking, 303 is a decoding table,
304 is a register unit that temporarily stores continuous zero run length;
305 is a register unit that temporarily stores non-zero coefficient values; 306
307 is a decoder unit that detects the number of coefficients from the zero run length of the register unit 304 and the non-zero coefficient value of the register unit 305;
308 is an intra-layer data amount setting unit that sets the data amount of each layer that can be divided into each block; 308 is an intra-layer data amount setting unit that selects one of the data amounts set in the intra-layer data amount setting unit 307;
309 is an adder that converts zero run lengths divided between layers into continuous zero run lengths; 310 is a register unit that temporarily stores non-zero coefficient values; 311 is a comparison unit that detects which layer the coefficient being processed belongs to, 312 is a register unit that temporarily stores the zero run length, 313 is a layer control unit, and 314 is a layer number setting that sets the number of layers to be converted into a sequential code. Department, 315
is a code table, 316 is a register section for temporarily storing codes, 317 is a register section for temporarily storing code lengths, 318
is a packing part that packs sequential codes,
319 is a code length control unit that controls the code length.

The mutual relationship and operation of the above-mentioned components will be explained below with reference to FIGS. 9 and 13. Figure 13(a)
, (b) and (c) show hierarchical encoded data obtained by encoding the block shown in FIG. 9 into three layers. here,
A case will be described in which the coefficients up to the B1 layer are converted into sequential codes. First, the intra-tier data amount setting unit 30
7, set the amount of data for each layer. In the case of the block shown in FIG. 9, the A layer is “5”, the B1 layer is “30”, and the B2 layer is “5”.
The hierarchy is set to "28". Further, the number of layers “2” is set in the number of layers setting section 314. In Figure 3, first, Figure 1
The hierarchical code data shown in 3(a) is sequentially input. In the address generation section 300, code data P-CODE-A-
An address is generated from 1 and given to the decoding table 303. P-CODE-A from the decoding table 303
Coefficient data for −1, zero run “3” and non-zero coefficient value “
The zero run "3" is taken into the register section 304, and the non-zero coefficient value "K1" is taken into the register section 305. In the comparator 308, the number of coefficients from the decoder 306 is "4".
” and the number of coefficients in the A layer, “5”, and inform the adding unit 309, the register unit 312, and the register unit 310 that the processing is in the A layer.
The register section 310 takes in the contents of 04 and the contents of the register section 305. Therefore, the code table 315 is given "3" from the register section 312 and "K1" from the register section 310 as addresses, and the code word S-CODE is given as an address.
1 and the code length are output and taken into the code word register section 316 and code length register 317. The packing unit 318 takes in the contents of the code word register 316, packs them, and outputs the contents when a predetermined amount is reached. Next, the address generation unit 300 generates the S-CODE under the control of the address generation control unit 301.
-Shift the data by the code length of A-1 and give the EOB code as an address to the decoding table. If EOB is detected here, the subtraction unit 308 subtracts the difference value “1” between the number of data in the A layer “5” of the intra-layer data amount setting unit 307 and the number of processed data of the A layer “4” of the hierarchy control unit 313. and stores it in the register section 312. In the address generation unit 300, EO
The data is shifted by the code length of B and temporarily stored in the memory 302. Next, the data of the B1 layer is sent to the address generator 30.
0, generates an address from P-CODE-B1-1, and gives the address to the decoding table 303. From the decoding table 303, coefficient data for P-CODE-B1-1, zero run "12" and non-zero coefficient value "K2" are obtained. Zero run “12” is a non-zero coefficient value “12” in the register section 304.
K1'' is taken into the register section 305. Comparator 30
8, the number of coefficients from the decoder 306 is “13” and B1
The number of coefficients in the layers, "30", is compared, and the addition unit 309, register unit 312, and register unit 310 are notified that the processing is for the B1 layer. The register section 312 is the register section 30
The contents of 4 and the contents of the register section 312 are added and "13" is taken into the register 312. The register section 310 takes in the contents "K2" of the register section 305. Therefore, the code table 315 contains the register section 312 as an address.
"13" is given from the register section 310 and "K2" is given from the register section 310, and the code word S-CODE2 and the code length are outputted and taken into the code word register section 316 and the code length register 317. The packing unit 318 takes in the contents of the code word register 316, packs them, and outputs the contents when a predetermined amount is reached. In this way, hierarchical code data is converted into sequential code. FIG. 13-4 shows data obtained by converting hierarchical code data up to the B1 layer into sequential codes.

[0033]

As is clear from the above embodiments, the present invention provides a data conversion device that encodes the number of consecutive zeros and non-zero coefficient values and manages them hierarchically. When converting code data into hierarchical code data, if it is not possible to divide the code data into layers at once, it can be divided into layers into any number of layers without increasing the hardware scale by dividing the data into layers multiple times.

(1)-(b) When converting sequential code data into hierarchical code data, by processing only the code word of the layer being processed by the packing means, it is possible to increase the speed without increasing the hardware size. Can be processed.

(2) When converting hierarchical code data into sequential code data, only the code word of the layer being processed is processed by the address generation means, thereby enabling high-speed processing without increasing the hardware scale.

(3)-(a) Efficient compression can be achieved by encoding with a code table that takes into consideration the frequency of occurrence of zero coefficients divided by layer division.

(3)-(b) By dividing each layer into layers, efficient compression is possible because encoding is performed using a code table that takes into account the frequency of occurrence of divided zero coefficients.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a data conversion device according to an embodiment of the first problem-solving means of the present invention.

FIG. 2 is a block diagram showing the configuration of a data conversion device according to an embodiment of the second problem-solving means of the present invention.

FIG. 3 is a block diagram showing the configuration of a data conversion device according to an embodiment of the third problem-solving means of the present invention.

[Figure 4] Schematic diagram showing how the block is divided into three layers

[Figure 5] Schematic diagram showing how the block is divided into layers from 2 to 3 layers

[Figure 6] Schematic diagram showing how sequential code data is divided into layers and converted into layered code data

[Figure 7] Schematic diagram showing how sequential code data is divided into two layers and further converted into three-layer code data.

[Figure 8] Schematic diagram showing the state of coefficient data in a block

[Figure 9] Schematic diagram showing the state of coefficient data in a block

FIG. 10 is a schematic diagram showing a state in which sequential code data of the block in FIG. 8 is converted into hierarchical code data.

[Figure 11] Schematic diagram showing the state of converting hierarchical code data to sequential code data

[Figure 12] Schematic diagram showing the arrangement of coefficients obtained by decoding the sequential code in Figure 11

[Figure 13] Schematic diagram showing the state of converting hierarchical code data to sequential code data

[Figure 14] Schematic diagram of hierarchically storing hierarchical code data

[Figure 15] Schematic diagram showing the characteristics of the transform coefficients of cosine transform

[Figure 16] Schematic diagram showing the order in which coefficients are scanned during sequential encoding

[Figure 17] Schematic diagram showing the state of non-zero coefficient data and zero coefficients in a block

FIG. 18 is a schematic diagram showing encoded data obtained by encoding the coefficients of the block in FIG. 17.

[Explanation of symbols]

100 address generation section 101 decoding table 102, 103 register section 104 decoder section 105 address generation control section 106, 107 intra-layer data amount setting section 108 subtraction section 109 subtraction section 110 register section 111 detection section 112 register section 113 layer control section 114 code Table 115 Demultiplexer section 116, 117 Code word register section 118, 119 Code length register section 120 Code length control section 121, 122 Packing section 123 Multiplexer section

Claims (5)

[Claims] Claim 1: Divide an image into blocks consisting of N×N pixels (N: natural number), perform orthogonal transformation on the data in the blocks, and then perform quantization, and calculate the number of zero coefficients and non-zero coefficients obtained. Data conversion that divides code data obtained by encoding coefficient values into hierarchical code data that divides the inside of a block into at least two or more subbands and can handle the transform coefficients divided into the subbands hierarchically. In the apparatus, a data amount setting means for setting the number of coefficients in each layer in the block consisting of N×N pixels, and a decoding means for obtaining the number of zero coefficients and the non-zero coefficient value from the input code data;
a detection means for detecting consecutive zero coefficients between layers; a subtraction means for calculating the number of zero coefficients belonging to each layer from consecutive zero coefficients between layers; and, when zero coefficients are continuous between layers, the calculation means. encode the number of zero coefficients and non-zero coefficient values obtained by the decoding means, and if the zero coefficients are not consecutive between layers, encode the number of zero coefficients and the non-zero coefficient values obtained by the decoding means. Converting an encoding means and a packing means for packing variable-length code data continuously generated in the order of layers in the encoding means for each layer into a layered code having M (M: natural number) layers. Prepare as much as possible, and when converting to layered data with more than M layers, convert to layered code data with M or less layers, and then convert undivided layer code data to M or less layers. A data conversion device that repeats the process of converting into a number of hierarchical code data to obtain a hierarchical code divided into an arbitrary number of layers. 2. Divide an image into blocks consisting of N×N pixels (N: natural number), perform orthogonal transformation on the data in the blocks, and then perform quantization, and calculate the number of zero coefficients and non-zero coefficients obtained. Data conversion that divides code data obtained by encoding coefficient values into hierarchical code data that divides the inside of a block into at least two or more subbands and can handle the transform coefficients divided into the subbands hierarchically. The apparatus includes a data amount setting means for setting the number of coefficients in each layer in the block consisting of N×N pixels, a packing means for packing at least one or more variable length code data, and a data amount setting means for setting the number of coefficients in each layer in the block consisting of N×N pixels, a packing device for packing at least one or more variable length code data, and a data amount setting means for setting the number of coefficients in each layer in the block consisting of N×N pixels, coded data storage means for temporarily storing coded data; only the codewords of the layer currently being processed are processed by the packing means; the codewords of the layer whose processing is being interrupted are temporarily stored in the coded data storage means;
and a control means for controlling the code word to be retrieved from the code data storage means in the middle of processing and processed by the packing means when the processing is restarted, and the block is divided into an arbitrary number of layers from the sequential code. A data conversion device that converts data into codes. Claim 3: A block is divided into at least two or more subbands, and the hierarchical code data obtained by hierarchically encoding the transform coefficients divided into subbands is converted into a sequential code up to a set layer. In the data conversion method, the data amount setting means sets the number of coefficients of each layer in the block consisting of N×N pixels, the layer number setting means sets the number of layers to be reproduced, and the layered encoded data is arranged in layer order. address generation means for extracting and unpacking the code data packed for each layer into variable-length code words, and generating an address for the code word;
a code data storage means for temporarily storing the code word being unpacked by the address generation means; and a code data storage means for temporarily storing the code word in the layer being processed; stored in a data storage means;
When the processing is restarted, a control means takes out the code word in the middle of processing from the code data storage means and controls the address generation means to process it; a decoding means for detecting consecutive zero coefficients between layers, a detection means for detecting consecutive zero coefficients between layers, an arithmetic means for calculating the number of consecutive zero coefficients between layers from zero coefficients belonging to each layer, and a decoding means for detecting consecutive zero coefficients between layers; If so, encode the number of zero coefficients and non-zero coefficient values obtained by the calculation means,
If the zero coefficients are not consecutive between layers, the block is divided into an arbitrary number of layers by an encoding means for encoding the number of zero coefficients obtained by the decoding means and the non-zero coefficient values. A data conversion device that converts hierarchical code data into sequential code data. 4. The data conversion device according to claim 1, wherein when converting sequential code data into hierarchical code data, the code table is changed depending on the number of layers to be divided and the number of coefficients in each layer. 5. The data conversion device according to claim 1, wherein when converting sequential code data into hierarchical code data, the code table is changed in each layer depending on the number of layers to be divided and the number of coefficients in each layer. .