JP2005012496A - Adaptive variable length encoder, adaptive variable length encoding method, and adaptive variable length encoding program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an adaptive variable length encoder in which the memory capacity of a counter for storing occurrence frequency can be reduced, and to provide an adaptive variable length encoding method and an adaptive variable length encoding program. <P>SOLUTION: The adaptive variable length encoder comprises a data storage meas 2 for storing input data, an encoding table 4 representing the relation of the input data and encoded data, a means 3 for encoding the input data with reference to the encoding table, a means 5 for dividing the input data into some groups, a means 6 for operating the occurrence frequency of each group of input data, and a means 7 for updating the encoding table based on the occurrence frequency of group data. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an adaptive variable length coding apparatus, adaptive variable length coding method, and adaptive variable length coding for generating compressed encoded data by adaptively changing the code length of data such as image data. Regarding the program.
[0002]
[Prior art]
[Patent Document 1] JP-A-8-316848
In general, when data such as image data is recorded on a recording medium, data compression processing is performed. There are various compression processing methods. As a representative method, for example, the data appearance frequency disclosed in Japanese Patent Laid-Open No. 8-316848 (Patent Document 1) is used. Huffman coding method (adaptive variable length coding method). In this method, data compression efficiency is statistically increased by using data with high appearance frequency as encoded data with a short bit length and data with low appearance frequency as encoded data with a long bit length. It is said. In this Huffman encoding, it is possible to calculate and encode in real time, but usually, an encoding table prepared in advance is often referred to. In the case of table reference, there are a method that keeps using one table and a method that selects from a plurality of tables, but one of the excellent methods is coding based on the result of counting the frequency of occurrence of encoding target data by value. There is an adaptive Huffman coding method that updates the coding table as appropriate. According to this, even if the generation tendency of the data to be encoded changes, a table corresponding to that can be prepared, so that further efficient compression can be expected.
[0004]
[Problems to be solved by the invention]
However, in this encoding method, the maximum appearance frequency can be counted for each data value from 0 to 255 if the size of the data value that can be taken by the data such as image data, for example, the image data is 8 bits. In the above example, 256 memories are required. When image data from an image sensor having a number of pixels of several million pixels is encoded, such as a recent digital camera, The memory capacity becomes extremely large. For example, if the number of pixels is 1 million pixels and the image data per pixel is 8 bits, the conventional encoding method requires 256 20-bit counters to create an encoding table. Although it is possible to limit the maximum count value and reduce the memory capacity of the counter to some extent, the encoding accuracy is still lowered if a counter having a certain amount of memory capacity is not provided.
[0005]
The present invention has been made in order to solve the above-mentioned problems in the conventional adaptive variable length coding system, and an adaptive variable length coding apparatus capable of reducing the memory capacity of a counter for storing the appearance frequency, An object of the present invention is to provide an adaptive variable length coding method and an adaptive variable length coding program.
[0006]
[Means for Solving the Problems]
In order to solve the above problem, the invention according to claim 1 is an adaptive variable length encoding device that encodes the data with reference to an encoding table created based on the appearance frequency of the data. A data storage means for storing the input data, an encoding table representing the relationship between the input data and the encoded data, and the input data stored in the data storage means is encoded with reference to the encoding table Encoding means; grouping means for dividing the input data into several groups; group data appearance frequency calculating means for calculating data appearance frequency for each group of the grouped input data; and the group data And coding table updating means for updating the coding table based on the appearance frequency.
[0007]
According to a second aspect of the present invention, in the adaptive variable length coding apparatus according to the first aspect, the grouping means groups the input data at predetermined intervals in the order of the data values. It is characterized by this.
[0008]
According to a third aspect of the present invention, in the adaptive variable length coding apparatus according to the first aspect, the grouping means analyzes the input data by data analysis means for analyzing the mode of the data in order of the magnitude of the data value. Grouping is performed at intervals determined based on the results.
[0009]
The invention according to claim 4 is the adaptive variable length coding device according to claim 1, wherein data appearance frequency predicting means for predicting the appearance frequency of each data of the input data based on the group data appearance frequency, Furthermore, it is characterized by having.
[0010]
According to a fifth aspect of the present invention, in the adaptive variable length coding apparatus according to the first aspect, the coding table updating means performs coding based on the group data appearance frequency by the group data appearance frequency calculating means at a predetermined tamming. The update table is updated.
[0011]
The invention according to claim 6 is the adaptive variable length coding apparatus according to claim 5, wherein the predetermined tamming is for each frame, for each line, or for each key frame.
[0012]
The invention according to claim 7 is the adaptive variable length coding device according to any one of claims 1 to 6, wherein the data of the coding table used when coding the input data is coded data. An encoding table data embedding unit for embedding is further provided.
[0013]
The invention according to claim 8 is an adaptive variable length encoding method that performs an encoding process of the data with reference to an encoding table created based on the appearance frequency of the data, the step of storing input data And dividing the stored input data into several groups, calculating a data appearance frequency for each group of the grouped input data, and based on the data appearance frequency for each group The method comprises a step of updating a coding table and a step of coding the stored input data with reference to the updated coding table.
[0014]
The invention according to claim 9 is the adaptive variable length encoding method according to claim 8, further comprising the step of embedding the data of the encoding table used when encoding the input data in the encoded data. It is a feature.
[0015]
The invention according to claim 10 is an adaptive variable-length encoding program for causing a computer to execute an encoding process of the data with reference to an encoding table created based on the appearance frequency of the data. A procedure for storing the input data, a procedure for dividing the stored input data into several groups, a procedure for calculating the data appearance frequency for each group of the grouped input data, and for each group The encoding table is updated based on the data appearance frequency and the stored input data is encoded with reference to the updated encoding table. .
[0016]
The invention according to claim 11 is the adaptive variable length encoding program according to claim 10, further comprising a procedure of embedding the data of the encoding table used when encoding the input data into the encoded data. It is a feature.
[0017]
DETAILED DESCRIPTION OF THE INVENTION
Next, embodiments will be described. FIG. 1 is a block diagram showing an embodiment of an adaptive variable length coding apparatus according to the present invention, and is also an explanatory diagram for explaining an embodiment of an adaptive variable length coding method. In addition, although this embodiment has shown the aspect which encodes image data, the data used as the object of an encoding are not restricted to image data. In FIG. 1, 1 is an image sensor that photoelectrically converts a subject image to generate an image signal, and 2 is data that stores image data obtained by A / D converting an image signal from the image sensor 1 by an A / D converter (not shown). Storage means 3 is an encoding means for encoding image data based on the appearance frequency of the data and outputs it as encoded data, and 4 is an encoding table used by the encoding means 3 indicating the relationship between the image data and the encoded data. An entropy coding table such as Huffman Langles, a lookup table, or the like is used. 5 is a grouping means for dividing the image data into several groups, 6 is a group data appearance frequency calculating means for calculating the data appearance frequency for each group of the data grouped by the grouping means 5, and 7 is group data. It is a coding table update means for updating the coding table based on the group data appearance frequency obtained by the appearance frequency calculation means 6.
[0018]
Next, the operation of the adaptive variable length coding apparatus configured as described above will be described. The image data generated by the image sensor 1 and A / D converted is temporarily stored in the data storage means 2. The image data stored in the data storage unit 2 is input to the encoding unit 3 and simultaneously to the grouping unit 5. At this time, the image data may not be temporarily stored in the data storage unit 2 but may be directly input to the encoding unit 3 and the grouping unit 5. Further, the input of the image data to the encoding means 3 and the grouping means 5 may be input in a shifted manner instead of simultaneously.
That is, the processing time in the grouping means 5 and the group data appearance frequency calculating means 6 is required, and there is a time lag until the encoding table is updated based on the calculation result and reflected in the encoding means 3. When the input to the grouping unit 5 precedes the input to the encoding unit 3, an encoding table optimum for input image data can be prepared in real time during the encoding process in the encoding unit 3. become. In order to delay the input to the encoding means 3 in this way, the image data may be input via the delay buffer.
[0019]
In the grouping means 5, the data values of the data to be encoded by the encoding means 3 are grouped under a specific condition. For example, assume that the input data is 8 bits and a value of 0 to 255 appears. For example, as shown in FIG. 2, the encoding target values of 0 to 15 are group 1 and the encoding target values of 16 to 31 are group 2. ,... Are encoded into groups 16 as in group 16, and the group data appearance frequency calculation means 6 counts the frequency of occurrence within each group.
[0020]
By counting the occurrence frequency by grouping in this way, the counter can be reduced to 1/16, and the memory capacity of the counter can be reduced. Of course, the information is reduced by this grouping, so the accuracy is lowered, but it is not a big problem in terms of the distribution of the occurrence frequency of general image data and moving image data.
[0021]
In the above grouping example, the input data is simply divided into 16 in order of the size of the data value. However, as a grouping technique with higher accuracy than this technique, there is the following technique. That is, the table can be updated more accurately and more accurately when the value having a higher appearance frequency is grouped into smaller groups. Therefore, in the case of a still image or a moving image, the spatial difference value or the frame difference value is more frequently generated as the value is usually closer to zero. Therefore, the encoding target values 0 to 1 are assigned to the groups 1 to 5. The encoding target values are grouped such that the encoding target values of groups 2 and 6 to 11 are the encoding target values of group 3,... 192 to 255. Thereby, grouping with relatively high accuracy can be performed. At this time, if a means for analyzing the characteristics of the occurrence frequency of the input data is separately provided, and the grouping interval is set based on the analysis result of the analysis means, it is more suitable for the input data. Table update is possible.
[0022]
In the above grouping method, the fixed number of encoding target values in each group has been described. However, it is also effective to allow the user to freely set the grouping conditions. In other words, for example, the number of groups is fixed at 16, and n1 is assigned to group 1, n2 is assigned to group 1,. The encoding target values n1 to n16 of each group can be configured to be changeable by a register.
[0023]
For example, even if the encoding target value is distributed around zero, for example, as shown in FIG. 3A, it is concentrated in the center of zero, and as shown in FIG. As shown in FIG. 3A, although the center is zero, the distribution may be gentle. Therefore, in the case of the distribution shown in FIG. However, in the case of the distribution shown in FIG. 3B, if the fine grouping centered on zero is performed, the influence on other gentle distribution areas becomes large, resulting in inappropriate grouping. Therefore, in such a case, it is preferable that the user can appropriately and arbitrarily change the group size.
[0024]
As described above, the data grouped by the grouping means 5 is counted for each group by the group data appearance frequency calculating means 6.
Then, the coding table updating means 7 compares the coding table determined based on the appearance frequency obtained by the group data appearance frequency calculating means 6 with the currently selected coding table, and based on the appearance frequency. If the new coding table determined in this step is different from the currently selected coding table, the coding means 3 uses the new coding table for the timing appropriate for updating the coding table. Select and update as a table. The encoding unit 3 performs an encoding process while referring to the updated encoding table.
[0025]
Various timings can be set as the update timing by the encoding table update means 7, for example, for each line, for each frame, for each key frame (frame for which encoding processing is performed without using information of other frames). Update at the timing. The update timing may be automatically determined and updated by the encoding table update means 7. For example, in the currently used encoding table, when the appearance frequency of the encoding target value of the group A to which the input data to which the shortest code is assigned belongs is lower than that of the other group B, the encoding table The updating means 7 is configured to detect this state and automatically update the coding table that assigns the shortest code to the input data belonging to the group B.
[0026]
In the above-described embodiment, the coding table corresponding to the data appearance frequency for each group of the data grouped by the grouping means 5 is shown. Since there is roughness without continuity, an improvement technique such as linear interpolation can be used as shown in FIG. 4 in order to reduce the roughness. That is, a separate data appearance frequency prediction means is provided, and the appearance frequency for each individual encoding target value is predicted by linear interpolation as indicated by the dotted line, and the table is updated in accordance with the predicted value. As a result, the roughness due to grouping is alleviated, and an accurate coding table can be reconstructed.
[0027]
In the above embodiment, when the encoded data is output by the encoding means 3, as shown in FIG. 5A, the encoding table used by the encoding means 3 in the encoded data string is displayed. The encoded data may be configured in a format for decoding by embedding data. In this case, an embedding function is provided in the encoding means. The timing of embedding the encoding table data is embedded in the encoded data string at the timing of updating the encoding table, that is, at the timing of line unit or frame unit, and the encoded data is configured. At this time, if data compressed by any means is used as the encoding table data to be embedded, the data size does not need to be increased unnecessarily.
[0028]
In the encoded data in which the encoding table data is embedded by the above-described embedding method, it is difficult to know where the encoding table data is inserted in terms of the data format. As described above, in a frame unit, a plurality of encoding table data used when encoding the encoded data of the frame is written together in the frame header, and the update point of the encoding table is set in the encoded data string. There is also a method of inserting only the information shown.
[0029]
Furthermore, as shown in FIG. 5C, a plurality of encoding table data are written together in the frame header, and at the same time, coordinate data for updating the encoding table is written. Also, it may be configured to output continuously without inserting.
[0030]
As described above, when the encoded data is configured in a format in which the encoding table information is embedded in the encoded data, decoding is performed based on the encoding table information and timing information embedded in the encoded data. Processing is performed.
[0031]
Encoding processing of data such as image data by the adaptive variable length encoding device according to the above embodiment is not limited to hardware processing, but software that causes each processing procedure by each of the above constituent means to be programmed and executed by a computer Of course, processing is also possible.
[0032]
【The invention's effect】
As described above based on the embodiments, according to the present invention, the input data is grouped into several groups, and the appearance frequency is calculated in units of groups. Therefore, the counter memory that stores the appearance frequency An adaptive variable length coding apparatus, an adaptive variable length coding method, and an adaptive variable length coding program capable of reducing capacity can be realized.
[Brief description of the drawings]
FIG. 1 is a block configuration diagram showing an embodiment of an adaptive variable length coding apparatus according to the present invention.
FIG. 2 is a diagram showing an example of grouping by the grouping unit in the embodiment shown in FIG. 1;
FIG. 3 is a diagram showing an occurrence frequency mode when encoding target values are distributed around zero.
FIG. 4 is a diagram illustrating an aspect of predicting appearance frequencies for individual encoding target values by linearly interpolating data appearance frequencies by grouping.
FIG. 5 is a diagram illustrating a mode in which encoding table data is embedded in encoded data output from an encoding unit.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Image pick-up element 2 Data storage means 3 Encoding means 4 Encoding table 5 Grouping means 6 Group data appearance frequency calculation means 7 Encoding table update means

Claims (11)

An adaptive variable-length encoding device that encodes the data with reference to an encoding table created based on the appearance frequency of data, the data storage means for storing input data, and the input data and code A coding table representing a relationship with the coded data, coding means for coding the input data stored in the data storage means with reference to the coding table, and dividing the input data into several groups Grouping means, group data appearance frequency calculating means for calculating the data appearance frequency for each group of the grouped input data, and encoding for updating the encoding table based on the group data appearance frequency And an adaptive variable-length coding apparatus comprising: a table updating unit; 2. The adaptive variable length coding apparatus according to claim 1, wherein the grouping means groups the input data at predetermined intervals in order of the magnitude of the data value. The said grouping means groups input data for every interval determined based on the analysis result by the data analysis means which analyzes the aspect of the said data in order of the magnitude | size of a data value. Adaptive variable length coding device. The adaptive variable length coding apparatus according to claim 1, further comprising data appearance frequency predicting means for predicting an appearance frequency of each data of the input data based on the group data appearance frequency. 2. The adaptive variable length coding apparatus according to claim 1, wherein the coding table updating unit updates the coding table based on a group data appearance frequency by the group data appearance frequency calculating unit at a predetermined tamming. . 6. The adaptive variable length coding apparatus according to claim 5, wherein the predetermined tamming is performed for each frame, for each line, or for each key frame. 7. The method according to claim 1, further comprising encoding table data embedding means that embeds data of an encoding table used when encoding the input data into encoded data. Adaptive variable length coding device. An adaptive variable length encoding method for performing an encoding process of the data with reference to an encoding table created based on the appearance frequency of data, the step of storing input data, and the stored input data Dividing the data into several groups, calculating the data appearance frequency for each group of the grouped input data, and updating the encoding table based on the data appearance frequency for each group; And encoding the stored input data with reference to the updated encoding table. An adaptive variable length encoding method comprising: 9. The adaptive variable length encoding method according to claim 8, further comprising a step of embedding data in an encoding table used when encoding input data in the encoded data. An adaptive variable-length encoding program for causing a computer to execute an encoding process of data by referring to an encoding table created based on the appearance frequency of data, and a procedure for storing input data And the procedure for dividing the stored input data into several groups, the procedure for calculating the data appearance frequency for each group of the grouped input data, and the data appearance frequency for each group An adaptive variable-length encoding program characterized by causing a procedure for updating an encoding table and a procedure for encoding the stored input data with reference to the updated encoding table. 11. The adaptive variable length coding program according to claim 10, further comprising a step of embedding coding table data used when coding input data into the coded data.

Images