JP4493552B2 - Data encoding apparatus, data decoding apparatus, data encoding method, data decoding method, and program - Google Patents

Description

  The present invention relates to a data encoding device, a data decoding device, a data encoding method, a data decoding method, and a program for encoding or decoding data based on an encoding table.

  In an imaging apparatus such as a digital camera, for example, in a single-lens reflex type model, imaging data called RAW data is recorded on a recording medium. This RAW data is simply converted into digital data after performing simple analog processing on the image signal obtained from the image sensor. By the way, in recent years, as the number of pixels of an image sensor of a digital camera is increasing, the data size of such RAW data is increasing. Accordingly, there are digital cameras that perform compression processing on RAW data to reduce the data size. However, it is assumed that the RAW data is retouched by the user, and it is desirable to perform lossless compression when performing compression processing.

The compression method is generally desired to have a high compression rate, but in the case of a compression method applied to an imaging device such as a digital camera, unlike the compression method applied to a single personal computer or the like, There are some more desirable conditions due to the restrictions on the specifications and the like, for example, as follows.
(1) Processing time is short (2) Buffer amount required for processing is small

  The condition (1) among these is a condition related to, for example, continuous shooting performance and battery consumption. Condition (2) is a condition related to the cost, size, weight, etc. of the digital camera.

  However, if both the condition (1) and the condition (2) are satisfied, the compression performance is likely to deteriorate.

  By the way, although there are various encoding methods, Huffman encoding used for image compression in JPEG or the like is one of well-known variable length encoding methods. When performing this Huffman coding, since it is possible to perform more efficient compression by using an appropriate coding table, there are various methods for selecting a coding table. Proposals have been made.

For example, Japanese Patent Laid-Open No. 9-65334 describes a technique for selecting an encoding table based on the ratio of zero coefficient signals. Therefore, it is considered that the technique described in the publication can achieve a high compression rate when the ratio of the zero coefficient signal is large, for example, in the case of a lossy compression method with quantization.
JP-A-9-65334

  However, in lossless compression, the ratio of the zero coefficient signal is often relatively small. Therefore, in the technique based only on the zero coefficient as described in JP-A-9-65334, the encoding table to be selected is not selected. It is difficult to determine with high accuracy. Further, in the technique for evaluating based on the ratio of the zero coefficient signal as described in the publication, since the coding efficiency (compression efficiency) with respect to a value other than the zero coefficient in the image block is not taken into consideration, the edge particularly When processing is performed on an image block including “”, the encoding efficiency may be extremely reduced.

  Therefore, not only lossy compression but also lossless compression, a compression method that can exhibit high compression performance while satisfying the above conditions (1) and (2) is desired.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a data encoding device, a data encoding method, and a program that have a short processing time, require a small amount of memory, and can exhibit high compression performance. It is said.

  It is another object of the present invention to provide a data decoding apparatus, a data decoding method, and a program that can decode the compressed data with a short processing time and a small required memory amount.

In order to achieve the above object, the data encoding apparatus according to the first invention detects the continuous number of the data whose data values are continuously within a predetermined range, and the continuous number of the data is predetermined. And a lancet counter for detecting the number of lancets that is the number of times that the continuation number of the data is detected to have reached the predetermined number by the run counter. And a minimum code length category determination that determines a minimum code length category that is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets detected by the lancet counter. means a minimum code length thus significant number of bits represented in the category determined by the minimum code length category determination means For a category representing the closest number significantly bits, to assign a shorter code length category codes, and the coding table creation means for creating a coding table showing the correspondence between the category and the category code, the coding table Encoding means for generating encoded data by associating the category code with the significant bit portion of the data based on the encoding table created by the creating means, and setting the plural types of the predetermined range The run counter detects the continuous number of the data in which the data value is continuously within the predetermined range for each of the plurality of predetermined ranges set, The detection is repeated every time the continuous number of the data reaches a predetermined number, and the lancet counter is The number of lancets, which is the number of times that the number of consecutive data reaches the predetermined number, is detected corresponding to each of the predetermined ranges of the plurality of types, and the minimum code length category determination is performed. A means for assigning a category code having a minimum code length among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. The minimum code length category is determined .

  The data encoding apparatus according to the second invention is the data encoding apparatus according to the first invention, wherein the encoding target data to be noticed is predicted based on the encoding target data read in the past, and the prediction data is calculated. A prediction error calculating means for generating prediction error data by taking a difference between the target encoding target data and the prediction data calculated by the prediction means. This is prediction error data generated by the prediction error calculation means.

A data encoding apparatus according to a third invention is the data encoding apparatus according to the first invention, wherein the minimum code length category determining means is configured such that the number of lancets detected by the lancet counter is equal to or greater than a predetermined number. The minimum code length category is determined so that the category code having the minimum code length is assigned to the category representing the smaller number of significant bits than when the number is less than the predetermined number.

A data encoding device according to a fourth invention is the data encoding device according to the third invention, wherein the predetermined range is a range in which an absolute value of the data is equal to or less than a predetermined threshold, and the predetermined threshold Can be set for a plurality of types, and the run counter has a continuous number of data for which the absolute value of the data is continuously equal to or less than the predetermined threshold for each of the set types of predetermined thresholds. And the detection is repeated every time the continuous number of the data reaches a predetermined number . The lancet counter detects that the continuous number of the data has reached the predetermined number by the run counter. the number of the lancet which is the number of times that is, in correspondence with each of the plurality of types of predetermined threshold is used to detect, the minimum code length category determination means, the lancet mosquito When the number of lancets detected corresponding to each of the plurality of types of predetermined threshold values is greater than or equal to a predetermined number, and the lowest threshold value that is greater than or equal to the predetermined number is high when it is low The minimum code length category is determined so that the category to which the category code of the minimum code length is assigned is a category representing a smaller number of significant bits.

A data encoding device according to a fifth aspect of the present invention is the data encoding device according to the first aspect of the present invention, wherein the information relating to the encoding table created by the encoding table creating means and the code generated by the encoding means The digitized data is output as a single file.

In the data encoding method according to the sixth aspect of the invention, the run counter detects the continuous number of the data whose data values are continuously within a predetermined range, and each time the continuous number of the data reaches the predetermined number. A first step of repeatedly performing the detection, a second step of detecting by the lancet counter the number of lancets, which is the number of times that the continuation number of the data is detected by the lance counter, and a minimum the code length category determination unit, according to the number of the detected lancet by the lancet counter, to determine the minimum code length category is a category to assign a minimum code length category code of the category representing the number of significant bits of the data a third step, thus the table the minimum code length categories determined by the minimum code length category determination means For a category representing the closest number significant bits of the number of significant bits, to allocate shorter code length category code, and a fourth step of creating a coding table showing the correspondence between the category and the category code, A fifth step of generating encoded data by associating the category code with the significant bit portion of the data based on the encoding table generated by the encoding table generating means , and the predetermined range A plurality of types can be set, and the run counter calculates, for each of the set types of predetermined ranges, the continuous number of the data in which the data values are continuously within the predetermined range. And the detection is repeated each time the number of consecutive data reaches a predetermined number. The lancet counter The number of lancets, which is the number of times that the number of consecutive data has been detected to reach the predetermined number, is detected corresponding to each of the plurality of predetermined ranges, and the minimum code length category The determining means assigns a category code having a minimum code length among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. This is a method for determining a minimum code length category that is a category .

According to a seventh aspect of the invention, there is provided a program for detecting a continuous number of data in which a data value continuously falls within a predetermined range and repeating the detection every time the continuous number of data reaches a predetermined number. A lancet counter for performing, a lancet counter for detecting the number of lancets that is the number of times that the continuation number of the data is detected to have reached the predetermined number by the run counter, and the lancet counter detected by the lancet counter depending on the number determined, the minimum code length category determination means for determining a minimum code length category is a category to assign a minimum code length category code of the category representing the number of significant bits of the data, by the minimum code length category determination means Table number significant bits close to the minimum code length thus significant number of bits represented in the category that is For a category, to assign a shorter code length category code, the coding table creation means for creating a coding table showing the correspondence between the category and the category code, the code created by the coding table creation means A program for functioning as an encoding means for generating encoded data by associating the category code with a significant bit portion of the data based on an encoding table, wherein a plurality of types of the predetermined range can be set The run counter detects, for each of a plurality of set predetermined ranges, the continuous number of the data whose data values are continuously within the predetermined range, and the data The detection is repeated every time the continuous number reaches a predetermined number, and the lancet counter is The number of lancets, which is the number of times that the continuous number of data reaches the predetermined number, is detected corresponding to each of the predetermined ranges of the plurality of types, and the minimum code length category determination A means for assigning a category code having a minimum code length among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. Is a program for determining a minimum code length category .

According to an eighth aspect of the present invention, the data decoding apparatus detects the number of consecutive data whose data values are continuously within a predetermined range, and repeats the detection every time the number of continuous data reaches a predetermined number. A lancet counter for performing, a lancet counter for detecting the number of lancets, which is the number of times that the continuation number of the data has been detected to reach the predetermined number by the run counter, and the lancet counter Minimum code length category determining means for determining a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets, and the minimum code length category determining means catheter representing the closest number significant bits Thus the number of significant bits represented the minimum code length categories determined by A coding table creation means for creating a coding table showing the correspondence between the category and the category code to allocate shorter code length category code from the more Li, the coding table created by said coding table creation means Encoding means for associating the category code with the significant bit portion of the data to generate encoded data, and the predetermined range can be set to a plurality of types, and the run counter is , For each of a plurality of set predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range is detected, and the continuous number of the data reaches the predetermined number The detection is repeated every time, and the lancet counter has reached the predetermined number of continuous data by the run counter. The number of detected lancets is detected in correspondence with each of the plurality of types of predetermined ranges, and the minimum code length category determination means uses the lancet counter to determine the plurality of types of predetermined ranges. In accordance with the number of lancets detected corresponding to each of the above, the minimum code length category, which is a category to which the category code of the minimum code length is assigned among the categories representing the number of significant bits of the data, is determined. A data decoding device for decoding encoded data generated by a featured data encoding device , wherein the code used for encoding is based on the information related to the generated encoding table. An encoding table reproducing means for reproducing the encoding table, and an encoding table reproduced by the encoding table reproducing means Decoding means for decoding the encoded data and reproducing the data.

In the data decoding method according to the ninth aspect of the invention, the run counter detects the continuous number of the data whose data values are continuously within a predetermined range, and each time the continuous number of the data reaches the predetermined number. A first step of repeatedly performing the detection, a second step of detecting by the lancet counter the number of lancets, which is the number of times that the continuation number of the data has been detected by the lance counter, and a minimum The code length category determining means determines a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets detected by the lancet counter . 3 and step, thus the table the minimum code length categories determined by the minimum code length category determination means More categories representing the closer number significant bits of the number of significant bits, and a fourth step of creating a coding table showing the correspondence between the category and the category code to allocate shorter code length category code, the coding table anda fifth step of generating coded data in association with the significant bit part of the category code and the data on the basis of the coding table created by the creating means, the predetermined range is a plurality of types set The run counter detects the continuous number of the data in which the data value is continuously within the predetermined range for each of the plurality of predetermined ranges set, The detection is repeated every time the number of consecutive data reaches a predetermined number, and the lancet counter is incremented by the run counter. Detecting the number of lancets, which is the number of times that it has been detected that the number of consecutive data has reached the predetermined number, corresponding to each of the predetermined ranges of the plurality of types, and determining the minimum code length category Is a category to which a category code having a minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets detected by the lancet counter corresponding to each of the plurality of types of predetermined ranges. A data decoding method for decoding encoded data generated by a data encoding method characterized in that a certain minimum code length category is determined , wherein the encoded table includes Based on such information, the encoding table used for encoding is reproduced, and the encoded data is reproduced based on the reproduced encoding table. This is a method of decoding data and reproducing data.

According to a tenth aspect of the invention, the computer detects the number of consecutive data whose data values are continuously within a predetermined range, and repeatedly performs the detection every time the number of consecutive data reaches the predetermined number. run counter, the number of the lancet consecutive number of the data by the run counter is detected lancet counter for detecting the number of the lancet which is the number of times that has been detected to have reached the predetermined number, by the lancet counter for determined by the minimum code length category determination unit, the minimum code length category determination means for determining a minimum code length category is a category to assign a minimum code length category code of the category representing the number of significant bits of the data in accordance with the It represents the number of significant bits close to the minimum code length thus significant number of bits represented in the category Encoding table generating means for generating an encoding table showing the correspondence between the category and the category code to allocate shorter code length category code from more categories, based on the coding table created by said coding table creation means The category code and a significant bit part of the data are associated with each other to generate encoded data, and a program for causing the predetermined range to be set in a plurality of types. The run counter detects, for each of a plurality of set predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range, and the continuous number of the data is The detection is repeated every time the predetermined number is reached, and the lancet counter is The number of lancets, which is the number of times that it has been detected that the continuous number has reached the predetermined number, is detected corresponding to each of the plurality of predetermined ranges, and the minimum code length category determining means includes: The minimum which is a category to which a category code having a minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. A program for causing a computer to execute a process of decoding encoded data generated by a program characterized by determining a code length category, the information relating to the generated encoding table Based on the above, the encoding table used for encoding is reproduced, and based on the reproduced encoding table, the above A program for causing a computer to execute a process of decoding encoded data and reproducing the data.

  According to the data encoding device, the data encoding method, and the program of the present invention, it is possible to exhibit high compression performance with a short processing time and a small required memory amount.

  In addition, according to the data decoding apparatus, data decoding method, and program of the present invention, the compressed data can be decoded with a short processing time and a small required memory amount.

  Embodiments of the present invention will be described below with reference to the drawings.

[Embodiment 1]
1 to 24 show Embodiment 1 of the present invention. FIG. 1 is a block diagram showing a configuration of a data encoding device, and FIG. 2 is a block diagram showing a configuration of the data decoding device.

  As shown in FIG. 1, the data encoding apparatus 1 includes a blocking unit 11, a prediction unit 12, a subtractor 13, an absolute value conversion unit 14, a flatness detection unit 15, a category determination unit 16, A Huffman table creation unit 17 and a Huffman encoding unit 18 are included. Of these, image data is input to the blocking unit 11, encoding table data is output from the Huffman table creation unit 17, and encoded data is output from the Huffman encoding unit 18, which is an encoding unit. It is like that. Details of the data encoding device 1 will be described later along the operation with reference to FIGS. 4 and 6 to 9.

  Further, the data compressed by the data encoding device 1 is decoded by a data decoding device 2 as shown in FIG. That is, the data decoding apparatus 2 includes a decoding table creation unit 21 as a coding table reproduction unit, a Huffman decoding unit 22 as a decoding unit, an adder 23, a memory 24, and a prediction unit 25. It is configured. Among these, the encoding table data is input to the decoding table creation unit 21, the encoding data is input to the Huffman decoding unit 22, and the image data is output from the adder 23. . The details of the data decoding device 2 will be described later along the operation with reference to FIG.

  In FIG. 1 and FIG. 2, the adjustment of the mutual operation timing between the blocks is performed by a timing adjustment circuit (not shown).

  Furthermore, the flatness detection unit 15 of the data encoding device 1 is configured as shown in FIG. FIG. 3 is a block diagram illustrating a configuration of the flatness detection unit.

  The flatness detection unit 15 includes a comparison unit 31, a run counter 32, a comparison unit 33, a lancet counter 34, and a comparison unit 35. Among these, the prediction error data and the threshold th are input to the comparison unit 31, the external count reset signal is input to the run counter 32, the threshold th_run is input to the comparison unit 33, and the lancet counter 34 is input to the lancet counter 34. An external count reset signal is input, and the threshold value th_run_set is input to the comparison unit 35.

  The details of the flatness detection unit 15 will also be described later along the operation.

  Next, FIG. 4 is a flowchart showing compression processing by the data encoding device 1, FIG. 6 is a flowchart showing processing for pixel units of flatness detection processing, and FIG. 7 is processing for line units of flatness detection processing. FIG. 8 is a flowchart showing processing for each block unit of flatness detection processing, FIG. 9 is a flowchart showing processing for creating a Huffman coding table, and FIG. 10 is a diagram showing how image data is divided into blocks. 11 is a diagram illustrating an example of pixel data of a block, FIG. 12 is a diagram illustrating an arrangement of a target pixel and neighboring pixels read before the target pixel, and FIG. 13 is a block prediction error obtained from the pixel data of FIG. FIG. 14 is a diagram showing data, FIG. 14 is a diagram showing prediction error data converted into absolute values obtained by converting the prediction error data in FIG. 13 into absolute values, and FIG. 15 is a diagram showing flatness detection. FIG. 16 is a chart showing the thresholds and setting values used and the minimum code length symbols corresponding to the thresholds, and FIG. 16 applies the thresholds and setting values of FIG. 15 to the absolute prediction error data of FIG. FIG. 17 is a chart showing the category pattern, FIG. 18 is a chart showing the relationship between the category corresponding to the minimum code length symbol and the code length, and FIG. 19 is the code length and category code. FIG. 20 is a diagram illustrating the structure of the Huffman code, FIG. 21 is a diagram illustrating an example of the prediction error data, and FIG. 22 is a diagram illustrating the Huffman code obtained by encoding the prediction error data in FIG. It is.

  When the compression process as shown in FIG. 4 is started, the blocking unit 11 first divides the input image data into blocks as shown in FIG. 10 (step S1). FIG. 10 shows an example in which image data of 32 × 56 pixels is divided into blocks of 8 × 8 pixels. As a result of the division, 7 × 4 blocks from B1 to B28 are obtained. Generated. Here, one image data is divided into 28 blocks, and one block is composed of 64 pixels. However, the present invention is not limited to this, so the number of blocks is more generally n, Assume that the number of pixels in a block is more generally m. Further, the present invention is not limited to dividing one image data into a plurality of pieces, and the entire image data may be processed as one block.

  Next, one block among a plurality of blocks obtained by division (specifically, a block in which pixel data of 8 bits (0 to 255) as shown in FIG. 11 is arranged) is processed. The prediction unit 12 predicts the pixel data of the pixel of interest in the block based on the pixel data already read in the block while sequentially reading out the pixel data of the pixels in the block (step S2). Here, as shown in FIG. 12, if the target pixel in the block is denoted by x, the neighboring pixel in the block read before the target pixel x is, for example, the pixel a adjacent to the left of the target pixel x. And a pixel b on the upper side of the target pixel x and a pixel c on the upper left side of the target pixel x. In this case, the pixel reading order is to sequentially read from the leftmost pixel to the rightmost pixel for a certain line in the block, and when the reading of the line is completed, the next line in the block is read in the same manner. However, the reading order is not limited to this.

Then, assuming that the pixel values of the pixels a, b, and c are Ra, Rb, and Rc, respectively, and the predicted value (predicted data) of the target pixel x is Px, the predicted value Px is the spatial correlation of such pixel data. Based on the above, for example, the prediction unit 12 serving as a prediction unit calculates using a prediction formula as shown in the following Formula 1.
[Equation 1]
Px = max (Ra, Rb) (when Rc <min (Ra, Rb))
Px = min (Ra, Rb) (when Rc> max (Ra, Rb))
Px = Ra + Rb-Rc (at other times)
Here, the symbol max (x, y) means that the smaller value of x and y is taken, and the symbol min (x, y) takes the lesser value of x and y. This is also used in the following.

In addition, although the example using the prediction formula suitable for edge detection like Formula 1 was demonstrated here, the applicable prediction formula is not restricted to this, Other various prediction formulas can also be used. Is possible. For example, the following formulas 2 to 8 are given as some examples of prediction formulas. First, Equation 2 sets the pixel value Ra of the pixel a adjacent to the left of the target pixel as the predicted value Px of the target pixel x. This is considered that a highly accurate prediction value can be obtained when the correlation between pixels in the horizontal direction is high.
[Equation 2]
Px = Ra

Next, Equation 3 sets the pixel value Rb of the pixel b above and below the target pixel as the predicted value Px of the target pixel x. This is considered that a highly accurate prediction value can be obtained when the correlation between pixels in the vertical direction is high.
[Equation 3]
Px = Rb

Further, Equation 4 sets the pixel value Rc of the pixel c adjacent to the upper left of the target pixel as the predicted value Px of the target pixel x. This is considered that a highly accurate prediction value can be obtained when the correlation between pixels in the diagonal direction from the upper left to the lower right is high.
[Equation 4]
Px = Rc

Then, Formula 5 adds the pixel value Ra of the pixel a adjacent to the left of the pixel of interest x and the pixel value Rb of the pixel b adjacent to the upper side of the pixel of interest x, and from there, The pixel value Rc is subtracted to obtain the predicted value Px of the target pixel x. This is considered that an accurate predicted value can be obtained when there is regularity in the rate of change of the pixel value in the diagonal direction from the upper left to the lower right.
[Equation 5]
Px = Ra + Rb-Rc

In addition, Expression 6 calculates the pixel value Ra of the pixel a adjacent to the pixel of interest x to the pixel value Ra of the pixel b adjacent to the pixel of interest x from the pixel value Rb of the pixel b adjacent to the pixel of interest x. The value obtained by subtracting and dividing by 2 is added to obtain the predicted value Px of the target pixel x. This is considered that when the change rate of the pixel value of the pixel in the horizontal direction has regularity, a highly accurate predicted value can be obtained.
[Equation 6]
Px = Ra + {(Rb−Rc) / 2}

Further, Equation 7 subtracts the pixel value Rc of the pixel c adjacent to the upper left of the pixel of interest x from the pixel value Ra of the pixel a adjacent to the left of the pixel of interest x to the pixel value Rb of the upper pixel b of the target pixel x. Then, the value divided by 2 is added to obtain the predicted value Px of the target pixel x. This is considered that when the change rate of the pixel value of the pixel in the vertical direction has regularity, an accurate predicted value can be obtained.
[Equation 7]
Px = Rb + {(Ra−Rc) / 2}

Further, Expression 8 is an average value of the pixel value Ra of the pixel a adjacent to the left of the target pixel x and the pixel value Rb of the pixel b adjacent to the upper side of the target pixel x as the predicted value Px of the target pixel x. is there. This is considered that a highly accurate prediction value can be obtained when the correlation with the surrounding pixels is high (for example, when it is relatively flat).
[Equation 8]
Px = (Ra + Rb) / 2

  In addition, each prediction formula as described above is not limited to using only one for one image data, and may be used in combination. That is, for example, even if a prediction formula as shown in Formula 2 is basically used, when the pixel at the left end on the line is the target pixel, there is no pixel on the left side, so Formula 3 A combination such as using the prediction formula shown in FIG. The pixel data of any one or more pixels is not limited to the above-described prediction formulas, and may be any pixel that is read before the target pixel and has a spatial distance close to the target pixel. It is possible to calculate a predicted value using

  Further, in the example of calculating the prediction error data as shown in FIG. 13 from the pixel data shown in FIG. 11, when calculating the prediction value before calculating the prediction error data, the pixel at the upper left corner in the block is the pixel. The value 127 of the dynamic range 0 to 255 is a predicted value, and the predicted value is calculated using Equation 3 for the leftmost pixel from the second line to the eighth line in the block, and for the other pixels The predicted value is calculated using Equation 2.

Next, as shown in Equation 9, the subtractor 13 serving as a prediction error calculation unit calculates the prediction error value pred_diff by subtracting the prediction value Px predicted in step S1 from the actual pixel value Rx of the target pixel. (Step S3).
[Equation 9]
pred_diff = Rx-Px

  The prediction error value pred_diff calculated in this way is data to be subjected to Huffman coding, and is used as an index for creating a Huffman coding table as well as used in the processing of Step S17 described later. In order to determine the flatness, it is used as described below. Note that when the prediction value Px is calculated as described above for the pixel data of the block as shown in FIG. 11, the prediction error value pred_diff as shown in FIG. 13 is obtained.

  Subsequently, the absolute value converting unit 14 converts the prediction error value pred_diff into an absolute value, and calculates an absolute value of the prediction error value abs_pred_diff (step S4). The absolute value conversion for converting the prediction error value pred_diff shown in FIG. 13 into the prediction error value abs_pred_diff shown in FIG. 14 is performed by removing the negative sign when it is a negative value.

Note that the absolute value conversion means is not limited to this, and may be performed as shown in the following Expression 10, for example.
[Equation 10]
abs_pred_diff = pred_diff (when pred_diff ≥ 0)
abs_pred_diff = abs (pred_diff) −1 (when pred_diff <0)
Here, abs () on the right side of the second expression indicates that it takes an absolute value.

  Since there are various means other than these for absolute value conversion, it is of course possible to apply such other means.

  Next, the flatness detection unit 15 performs flatness detection 2 based on the comparison of the flatness detection 1 process based on the comparison with the first threshold (step S5) and the second threshold. The process for the pixel unit of the process (step S6) and the process for the pixel unit of the process of flatness detection 3 based on the comparison with the third threshold value (step S7) are performed.

  The flatness detection 1 process in step S5, the flatness detection 2 process in step S6, and the flatness detection 3 process in step S7 are the same except that the value of the threshold th indicating a predetermined range is different. Since these processes are performed, these processes will be described together with reference to FIG. Further, as shown in FIG. 15, the specific threshold th is 7 (threshold th1) in the flatness detection 1 process in step S5, and 31 (threshold th2 here) in the flatness detection 2 process in step S6. In the process of flatness detection 3 in step S7, 63 (threshold value th3, where th3> th2 is satisfied) is used. As a threshold th_run indicating a predetermined run length as described later, specifically, 4 is commonly used in step S5, step S6, and step S7.

  When the processing for the pixel unit of the flatness detection processing is started, first, the comparison unit 31 determines whether or not the predicted error value abs_pred_diff converted into an absolute value is equal to or smaller than a predetermined threshold th (step S31). .

  Here, if it is equal to or less than the threshold th, the run counter 32 counts up the run counter run_count provided therein (here, the count up is indicated by the symbol “run_count + 1”) (step S32). . The run counter run_count is a counter indicating how many pixels the prediction error value abs_pred_diff converted into an absolute value that is equal to or less than the threshold th is continuous.

  Then, the comparison unit 33 determines whether or not the run counter run_count is equal to or greater than a threshold th_run indicating a predetermined run length (step S33).

  Here, when the value is equal to or greater than the threshold th_run, the lancet counter 34 counts up the lancet counter run_set_count (again, the count-up is indicated by the symbol “run_set_count + 1” as described above) (step S34). ). The lancet counter run_set_count is a counter indicating how many sets (lancets) in which the prediction error value abs_pred_diff whose absolute value is equal to or less than the threshold th is the threshold th_run are detected in the block.

  Thereafter, the run counter 32 initializes the run counter run_count (that is, stores “0”) in response to the result of step S34 (step S35).

  On the other hand, if it is determined in step S31 that the absolute value of the prediction error value abs_pred_diff is larger than the predetermined threshold th, the run counter 32 initializes the run counter run_count (step S36).

  In this way, when step S35 or step S36 is completed, or when it is determined in step S33 that the run counter run_count is less than the threshold th_run, the process returns to the process shown in FIG.

  As described above, when the processing of step S5, step S6, and step S7 is completed, it is subsequently determined whether or not the processing for one line in the block has been completed (step S8). If the processing for one line is not completed, the process returns to step S2 to perform the above-described processing for the next pixel on the same line.

  If it is determined that the processing for one line has been completed, the flatness detection unit 15 performs processing for each line unit (step S9) of the flatness detection 1 processing corresponding to step S5 described above. Processing for the line unit of the flatness detection 2 process corresponding to step S6 described above (step S10), processing for the line unit of the flatness detection 3 process corresponding to step S7 described above (step S11), I do.

  In addition, since the process of step S9, the process of step S10, and the process of step S11 perform the same process fundamentally, these processes are demonstrated collectively with reference to FIG.

  When this process is started, the run counter 32 receives an external count reset signal input from an external control means (not shown), etc., and resets an internal run counter run_count (step S41). Return to the indicated process.

  When the processing in step S9, step S10, and step S11 is completed in this way, subsequently, whether or not the processing for m pixels is completed, that is, whether or not the processing for one focused block is completed. Determination is made (step S12). If the process for one block has not been completed, the process returns to step S2 to perform the process as described above for the next line in the same block.

  On the other hand, if it is determined that the processing for one block has been completed, the flatness detection unit 15 performs processing for each block unit of the flatness detection 1 processing corresponding to steps S5 and S9 described above (steps). S13), the block unit processing (step S14) of the flatness detection 2 processing corresponding to steps S6 and S10 described above, and the flatness detection 3 processing block corresponding to steps S7 and S11 described above. Processing for unit (step S15) is performed.

  In addition, since the process of step S13, the process of step S14, and the process of step S15 perform the same process fundamentally, these processes are demonstrated collectively with reference to FIG. In this process, a threshold th_run_set indicating a predetermined number of lancets is used. As a specific numerical value, 4 is used as shown in FIG.

  When this process is started, the comparison unit 35 determines whether or not the lancet counter run_set_count counted by the lancet counter 34 is equal to or greater than the above-described threshold th_run_set (step S51).

  Here, if the comparison unit 35 determines that the threshold value is equal to or greater than the threshold th_run_set, the comparison unit 35 substitutes and outputs 1 for the run flag run_flag indicating the flatness detection result (step S52), and determines that the threshold value is less than the threshold th_run_set. In this case, 0 is assigned to the run flag run_flag indicating the flatness detection result and output (step S53). Here, based on the threshold value and the set value as shown in FIG. 15, the determination result of the prediction error value abs_pred_diff converted into the absolute value as shown in FIG. 14 is as shown in FIG. When the processing for one block is completed, the lancet counter run_set_count by flatness detection 1 is 0, the lancet counter run_set_count by flatness detection 2 is 3, and the lancet counter run_set_count by flatness detection 3 is 5. Since the threshold value th_run_set is set to 4 as described above, the run flag run_flag based on the flatness detection 1 is 0, the run flag run_flag based on the flatness detection 2 is 0, and the flatness detection 3 is performed, as shown in FIG. The run flag run_flag is 1.

  When the process of step S52 or step S53 is thus completed, the process returns to the process shown in FIG.

  As described above, after the processes of step S13, step S14, and step S15 are completed, the run flag run_flag output by the process of step S13, the run flag run_flag output by the process of step S14, and the process of step S15 15 determines which of the minimum code length symbols as shown in FIG. 15 is to be used by the category determination unit 16 as the minimum code length category determination means based on the run flag run_flag output by Huffman as the encoding table creation means. The table creation unit 17 creates a Huffman coding table according to the determined category (step S16). Here, the category determination unit 16 uses the minimum code length symbol corresponding to the smallest threshold th at which the run flag run_flag is 1, and in the example illustrated in FIGS. 15 and 16, the run flag run_flag is set to 1. Since only when the threshold value is th3, 6 is set as the minimum code length symbol. In addition, when all the run flags run_flag related to the thresholds th1 to th3 are 0, it is conceivable that, for example, 7 (or 8) is set as the minimum code length symbol, although not shown.

  Details of the Huffman coding table creation process will be described with reference to FIG.

  When this process is started, first, the variable i is initialized (that is, “0” is stored) (step S61). Here, the variable i indicates the category code to be assigned to which category code when the category codes as shown in FIG. 19 are from 0 to 8 from top to bottom. It is a variable. Note that since an 8-bit value is handled here, the variable i changes in the range of 0 ≦ i ≦ 8. More generally, however, when a k (k is an integer of 1 or more) bit value is handled. , The variable i changes in the range of 0 ≦ i ≦ k (this k is referred to as the maximum category value max_category in the following, so the range in which the variable i changes is written as 0 ≦ i ≦ max_category. Can also).

  Next, it is determined whether or not the variable i is equal to 0 (step S62). If the variable i is not equal to 0, the previous limit flag limit_flag [i-1] is not equal to 0 (here, FIG. 9). The symbol “! =” Indicates that they are not equal) (step S63). Further, the limit flag limit_flag is an i-th category pattern category_pattern [i] in the category pattern as shown in FIG. 17 in a category (category having the minimum code length) category_min to which a category code of the minimum code length is assigned as described later. ] Is within the range of possible category values (0 or more and max_category or less) (the limit flag limit_flag is “0” as will be described later), or the upper limit of possible category values ( It is a maximum category value max_category to be described later, and exceeds the max_category = 8 when handling an 8-bit value as shown in this example (the limit flag limit_flag is “1” as described later), Or it is below the lower limit of the possible category value (becomes “0”) (as described later, the limit flag limit_flag is “−1”), Is a flag for determining.

  Here, if the limit flag limit_flag [i-1] is not 0, it means that the category_min + category_pattern has already become larger than the maximum category value max_category or has become smaller than 0 before or before the previous time. For the sake of meaning, the same value as the previous limit flag limit_flag [i-1] is substituted for the limit flag limit_flag [i] of interest (here, the symbol “=” in FIG. 9 represents substitution) (The same applies hereinafter) (step S64).

  If the variable i is 0 in step S62 or if the previous limit flag limit_flag [i-1] is 0 in step S63, the i-th category pattern is included in the category category_min that is the minimum code length. It is determined whether the sum of category_pattern [i] is larger than the maximum category value max_category (step S65). Here, the category category_min that is the minimum code length is predetermined as a minimum code length symbol within the range of 0 ≦ category_min ≦ max_category, for example, as shown in FIGS. 15 and 18, according to the value of the threshold th. . In the examples shown in these figures, a minimum code length symbol of 3 for the threshold th1, a minimum code length symbol of 5 for the threshold th2, and a minimum code length symbol of 6 for the threshold th3, respectively. Is given.

  If it is determined in step S65 that the value is large, 1 is substituted into the limit flag limit_flag [i] of interest (step S66).

  On the other hand, if it is determined in step S65 that it is less than or equal to the maximum category value max_category, the value obtained by adding the i-th category pattern category_pattern [i] to the category category_min that is the minimum code length is less than zero. Is determined (step S67).

  Here, if it is less than 0, −1 is assigned to the limit flag limit_flag [i] to be noticed (step S68), and if it is 0 or more, 0 is assigned to the limit flag limit_flag [i] to be noticed. (Step S69).

  When any of the above-described processing of step S64, step S66, step S68, or step S69 is completed, the limit flag limit_flag [i] is determined (step S70).

  When it is determined that the limit flag limit_flag [i] is 0 (that is, when 0 ≦ (category_min + category_pattern [i]) ≦ max_category), the i th category pattern in the category category_min that is the minimum code length The sum of category_pattern [i] is set as the i-th category category [i] (step S71).

  If it is determined in step S70 that the limit flag limit_flag [i] is 1 (that is, if max_category <(category_min + category_pattern [i]) even once), the 0th category category [0] ] To the (i−1) th category category [i−1] minus 1 is defined as the i th category category [i] (step S72).

  Furthermore, if it is determined in step S70 that the limit flag limit_flag [i] is −1 (that is, if (category_min + category_pattern [i]) <0 even once), the 0th category category [ 0] to (i-1) th category category [i-1] plus 1 is defined as i-th category category [i] (step S73).

  When any of the processes from step S71 to S73 is completed, it is determined whether or not the variable i is less than the maximum category value max_category (step S74). If it is less than the maximum category value max_category, i is incremented ( In FIG. 9, (represented by the symbol “++”) (step S75), the process returns to step S62 to repeat the above-described processing.

  If it is determined in step S74 that the variable i is greater than or equal to the maximum category value max_category, the process returns to the process shown in FIG. 4 from the Huffman coding table creation process.

  If the process as shown in FIG. 9 is briefly described, when a category representing the code length of a significant bit is assigned to the category code as shown in FIG. 19, the category category_min that is the minimum code length is assigned. First, the category code (“00” in the example shown in FIG. 19) of the minimum code length (code length 2 in the example shown in FIG. 19) is given, and then the category category_min that becomes the minimum code length is shown in FIG. The category patterns category_pattern are added in order, and the category codes shown in FIG. 19 are assigned in order. At this time, if (category_min + category_pattern) exceeds the maximum category value max_category, the remaining category codes are assigned to descend from the largest category among the categories that have not yet been assigned. In such a case, the remaining category codes are assigned in ascending order from the smallest category among the categories not yet assigned.

  As a specific example of processing, when the minimum code length symbol is 3, first, category code “00” is assigned to category “3” indicating that the code length of significant bits is 3, and then “3” ”Is assigned a category code“ 010 ”to a category“ 4 ”obtained by adding a category pattern“ 1 ”, and a category code“ 011 ”is assigned to a category“ 2 ”obtained by adding a category pattern“ −1 ”to“ 3 ”. The category code “100” is assigned to the category “5” obtained by adding the category pattern “2” to the “3”, the category code “101” is assigned to the category “1” obtained by adding the category pattern “−2” to the “3”, A category code “110” is assigned to a category “6” obtained by adding a category pattern “3” to “3”, and a category “6” obtained by adding a category pattern “−3” to “3”. ”Is assigned a category code“ 1110 ”, a category pattern“ 4 ”is added to“ 3 ”, a category code“ 11110 ”is assigned to“ 7 ”, and a category pattern“ −4 ”is added to“ 3 ”. Since the corresponding category does not exist with “1”, the category code “111110” is assigned to the category “8” obtained by adding 1 to the maximum category “7” that has already appeared, and an 8-bit value is handled here. Since the assignment to all the categories from “0” to “8” necessary for the process is performed, an operation of terminating the processing is performed.

  For example, the same processing is performed when the minimum code length symbol is 6, but when i = 5, adding “6” to the category pattern “3” results in “9” and the corresponding category. Therefore, the category code “110” is assigned to the category “3” obtained by subtracting 1 from the minimum category “4” that has already appeared. Thereafter, the category “2”, the category “1”, and the category “0” are allocated. In addition, an operation of sequentially assigning category codes is performed.

  In the above-described example, since the minimum code length symbol “6” corresponding to the threshold th3 is set, the minimum code length symbol in FIG. 18 is 6 with respect to the category code shown in FIG. Each category will be assigned.

  Returning to the description of FIG. 4, when the processing of step S <b> 16 is completed, the Huffman encoding unit 18 as the encoding unit is calculated in step S <b> 3 by using the encoding table generated by the Huffman table generating unit 17. Huffman coding of the prediction error value pred_diff is performed (step S17).

  As is well known, the Huffman code is composed of a category code and significant bits (here, significant bits of prediction error data) (see FIG. 20). The category code is a code indicating the code length of the significant bit part. For example, 9-bit prediction error data “000001011” (decimal number “11”) as shown in FIG. 21 is obtained. Then, since the significant bits are for the lower 4 bits, the category is “4”. As described above, the category code assigned to the category 4 when the minimum code length symbol is 6 is “101” with reference to FIGS. 18 and 19, so the Huffman encoded data is As shown in FIG. 22, “1011011” is obtained. This encoded data “1011011” has 7 bits, and it can be seen that the data is compressed.

  Thus, when the Huffman encoding process is completed for each pixel data in one block, whether or not the process has been completed for n blocks, that is, the process of the entire image data to be processed is completed. Is determined (step S18). If the processing of the entire image data has not been completed, various parameters used for processing in units of blocks are reset as necessary, and the process returns to step S2 to return to the next in the same image data. The above-described processing is performed for this block. In addition, when the processing of the entire image data has been completed, this compression processing ends.

  The compressed image data is filed based on a predetermined file format in order to be recorded on a recording medium or the like via a file system or the like. FIG. 23 is a diagram illustrating an example of a file format for recording compressed image data.

  The compressed image file includes encoded data, and a header and a tag provided along with the encoded data.

  The Huffman coding table created for each block is recorded in, for example, a tag portion when generating a compressed image file.

  Here, FIG. 24 is a diagram showing a tag structure in the file format.

  The tag has a tag ID indicating the type of tag, a type indicating the type of the value recorded in the tag, a count indicating how many values of the type indicated by this type, and a value in the tag And an offset to a value indicating an address in which is stored.

  And in order to show the Huffman encoding table used with respect to a block, the following information is stored as tag information, for example.

  First, in the tag ID, for example, information that substantially indicates that “the value is stored in the Huffman encoding table for each block” is stored.

  Next, as a type, a type indicating the amount of data (specific examples include “char” indicating 1-byte data and “int” indicating an integer type as used in the programming language C). Is stored. Here, the type of data size required to store the Huffman coding table is stored.

  Furthermore, as the count, for example, the number of blocks is stored.

  As the offset, for example, the relative data movement amount from the head of the tag ID to the address where the value is stored, or the data movement from the head of the tag area shown in FIG. 23 to the address where the value is stored. Quantity, etc. are stored.

  Thereby, at the time of decoding, it is possible to read the Huffman coding table corresponding to the block and perform the expansion process.

  In the above-described example, three types of threshold values th1 to th3 are used, and the minimum code length when three types of minimum code length symbols corresponding to each threshold value and run flags run_flag corresponding to all threshold values are 0 is used. Since there are a total of four types of symbols, one type, there are four types of Huffman coding tables to be created. Therefore, it is not advantageous in terms of data amount to record the Huffman coding table as many as the number of blocks corresponding to each block. Therefore, for example, all the created Huffman coding tables are first recorded in tags, and appropriate codes are assigned to the Huffman coding tables (for example, corresponding to the minimum code length symbol “3” of the threshold th1). The code “0” in the Huffman coding table, the code “1” in the Huffman coding table corresponding to the minimum code length symbol “5” of the threshold th2, and the Huffman code corresponding to the minimum code length symbol “6” in the threshold th3. Code “2” in the encoding table, code “3” in the Huffman encoding table corresponding to the minimum code length symbol “7” (or “8”) when the run flag run_flag corresponding to all threshold values is 0, If any one of these codes “0” to “3” is recorded for each block in the tag, for each block, etc. To requires only data of bits, it is possible to reduce the data size. However, in practice, data is often stored in units of 1 byte. Therefore, in the above-described example, for example, it is also possible to store data for 4 blocks collectively in 1 byte.

  Thus, the Huffman coding table for each block, or the information indicating the types of all Huffman coding tables and the Huffman coding tables for each block, stored as a set of values corresponding to each block, is described above. Encoding table data (data output from the Huffman table creation unit 17 and data input to the decoding table creation unit 21).

  Next, FIG. 5 is a flowchart showing the decompression process performed by the data decoding apparatus 2.

  When this process starts, first, the decoding table creation unit 21 reads the encoding table data from the image file, reproduces the Huffman encoding table corresponding to each block, and outputs it to the Huffman decoding unit 22 (step S21). ).

  Subsequently, the Huffman decoding unit 22 reads the encoded data from the image file and decodes the prediction error value pred_diff for each pixel in the block based on the Huffman encoding table (step S22). .

  Then, the prediction unit 25 calculates the predicted value Px based on the already calculated pixel value stored in the memory 24 as described in step S2 of FIG. 4 (step S23).

Next, based on the prediction error value pred_diff decoded in step S22 and the prediction value Px calculated in step S23, a pixel value Rx is calculated as shown in the following equation 11 (step S24). .
[Equation 11]
Rx = Px + pred_diff

  Thereafter, it is determined whether or not the processing for one line in the block has been completed (step S25). If it has not been completed, the process returns to step S22 to perform the processing as described above for the next pixel in the same line. Repeat this step.

  If it is determined in step S25 that the processing for one line in the block has been completed, then whether or not the processing for m pixels has been completed, that is, the processing for one block has been completed. (Step S26), if not completed, the process returns to step S22, and the above-described processing is repeated for the next line in the same block.

  On the other hand, if it is determined in step S26 that the processing for one block has been completed, whether or not the processing has been completed for all n blocks, that is, the processing of the entire image data has been completed. (Step S27), and if not completed, the process returns to step S21 to read the encoding table data related to the next block and perform the above-described processing.

  In this way, when it is determined in step S27 that the processing of the entire image data has been completed, this expansion processing is ended.

  In the above description, image data is taken as an example of data to be processed. However, data that can be processed by the data encoding device 1 or the data decoding device 2 is not limited to this. .

  In the above description, processing is performed by the data encoding device 1 and the data decoding device 2, but processing may be performed by applying a data encoding method or a data decoding method to an existing arithmetic unit or the like. I do not care. Or you may make it make it process with a computer by the program for performing the process equivalent to the data encoding apparatus 1 or the data decoding apparatus 2. FIG.

  Furthermore, in the above description, the prediction error data is converted into an absolute value and then compared with a predetermined threshold value to detect the flatness, but the present invention is not limited to this. For example, the flatness may be detected according to whether prediction error data that has not been converted to an absolute value falls within a predetermined range.

  In the above description, the data to be subjected to Huffman coding and the flatness detection target is the prediction error data. However, the present invention is not limited to this, and other various data can be targeted. It is.

  In addition, in the above description, the Huffman coding table is taken as an example of the coding table to be created. However, the present invention is not limited to this, and coding tables in other coding methods can be created according to flatness. It is.

  According to the first embodiment, since the flatness is detected in units of blocks and the Huffman coding table is generated according to the flatness, the Huffman coding table suitable for the detected flatness is used. Huffman coding can be performed, and high compression performance can be exhibited.

  And, because the number of lancets in which a predetermined number of prediction errors converted into absolute values equal to or less than a predetermined threshold value are detected, it is determined that the lancet is flat when the lancet is equal to or greater than the predetermined number. In addition, flatness can be detected at high speed in a short time without requiring complicated processing.

  Furthermore, since a plurality of predetermined threshold values to be compared with the prediction error converted to the absolute value are prepared, the flatness can be detected finely.

  In addition, since it is only necessary to count by a run counter or a lancet counter when detecting flatness, an increase in the amount of memory required when creating a Huffman coding table can be substantially suppressed, saving It can be configured with a memory.

  Since the Huffman coding table is adaptively created, a storage means for storing a plurality of Huffman coding tables in advance is not necessary, and the memory can be further reduced.

  In addition, the compressed data can be decoded with a short processing time and a small necessary memory amount.

  It should be noted that the present invention is not limited to the above-described embodiments, and various modifications and applications can be made without departing from the spirit of the invention.

  The present invention can be suitably used for a data encoding device, a data decoding device, a data encoding method, a data decoding method, and a program for encoding or decoding data based on an encoding table.

1 is a block diagram showing a configuration of a data encoding device according to Embodiment 1 of the present invention. The block diagram which shows the structure of the data decoding apparatus in the said Embodiment 1. FIG. The block diagram which shows the structure of the flatness detection part in the said Embodiment 1. FIG. 6 is a flowchart showing compression processing by the data encoding apparatus according to the first embodiment. 7 is a flowchart showing decompression processing by the data decoding apparatus according to the first embodiment. 5 is a flowchart showing processing for pixel units in flatness detection processing in the first embodiment. 5 is a flowchart showing processing for line units in flatness detection processing in the first embodiment. 5 is a flowchart showing processing for a block unit of flatness detection processing in the first embodiment. 5 is a flowchart showing processing for creating a Huffman coding table in the first embodiment. The figure which shows a mode that image data is divided | segmented into a block in the said Embodiment 1. FIG. FIG. 3 is a diagram illustrating an example of pixel data of a block in the first embodiment. The figure which shows arrangement | positioning with the attention pixel in the said Embodiment 1, and the neighboring pixel read before an attention pixel. The figure which shows the prediction error data of the block obtained from the pixel data of the said FIG. The figure which shows the prediction error data converted into the absolute value obtained by converting the prediction error data of the said FIG. 13 into an absolute value. FIG. 5 is a chart showing thresholds and setting values used in flatness detection according to Embodiment 1 and minimum code length symbols corresponding to the thresholds. FIG. The chart which shows the result of the flatness detection obtained by applying each threshold value and setting value of FIG. 15 to the absolute value prediction error data of FIG. The chart which shows the category pattern in the said Embodiment 1. FIG. FIG. 3 is a chart showing a relationship between a category corresponding to a minimum code length symbol and a code length in the first embodiment. The chart which shows a response | compatibility with code length and a category code in the said Embodiment 1. FIG. The figure which shows the structure of the Huffman code | symbol in the said Embodiment 1. FIG. The figure which shows an example of the prediction error data in the said Embodiment 1. FIG. The figure which shows the Huffman code | cord | chord obtained by encoding the prediction error data of the said FIG. FIG. 3 is a diagram showing an example of a file format for recording image data compressed in the first embodiment. The figure which shows the structure of the tag in the file format of the said Embodiment 1. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Data encoding apparatus 2 ... Data decoding apparatus 11 ... Blocking part 12 ... Prediction part (prediction means)
13: Subtracter (prediction error calculation means)
DESCRIPTION OF SYMBOLS 14 ... Absolute value conversion part 15 ... Flatness detection part 16 ... Category determination part (minimum code length category determination means)
17 ... Huffman table creation unit (encoding table creation means)
18 ... Huffman encoding unit (encoding means)
21 ... Decoding table creation unit (encoding table reproduction means)
22 ... Huffman decoding unit (decoding means)
DESCRIPTION OF SYMBOLS 23 ... Adder 24 ... Memory 25 ... Prediction part 31 ... Comparison part 32 ... Run counter 33 ... Comparison part 34 ... Lancet counter 35 ... Comparison part

Claims (10)

A run counter for detecting the continuous number of the data whose value continuously falls within a predetermined range, and repeatedly performing the detection every time the continuous number of the data reaches a predetermined number ;
A lancet counter for detecting the number of lancets that is the number of times that the continuation number of the data has been detected to have reached the predetermined number by the lance counter;
Minimum code length category determining means for determining a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets detected by the lancet counter. ,
For a category representing the number of significant bits close to the significant number of bits represented depending on the minimum code length categories determined by the minimum code length category determination means, to allocate shorter code length category codes, category and category code A coding table creating means for creating a coding table indicating the correspondence relationship with
Encoding means for generating encoded data by associating the category code with the significant bit portion of the data based on the encoding table created by the encoding table creating means;
Equipped with,
A plurality of types of the predetermined range can be set,
The run counter detects, for each of a plurality of set predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range, and the continuous number of the data is The detection is repeated every time a predetermined number is reached,
The lancet counter detects the number of lancets corresponding to each of the predetermined ranges of the plurality of types, the number of times that the continuation number of the data is detected by the run counter to reach the predetermined number. And
The minimum code length category determining means includes a minimum code length among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. Determines the minimum code length category that is the category to which the category code is assigned
A data encoding apparatus characterized by that. Prediction means for predicting the encoding target data of interest based on the encoding target data read in the past and calculating prediction data;
Prediction error calculation means for generating prediction error data by taking the difference between the target encoding target data and the prediction data calculated by the prediction means;
Further comprising
The data encoding apparatus according to claim 1, wherein the data is prediction error data generated by the prediction error calculation means. When the number of lancets detected by the lancet counter is greater than or equal to a predetermined number, the minimum code length category determining means determines that the minimum code length category determining means has a minimum number of categories representing a significant number of bits as compared to when the number is less than the predetermined number. The data encoding apparatus according to claim 1, wherein a minimum code length category is determined so as to assign a code code of a code length . The predetermined range is a range in which the absolute value of the data is equal to or less than a predetermined threshold, and the predetermined threshold can be set in a plurality of types.
The run counter detects, for each of a plurality of predetermined threshold values set, the continuous number of the data whose absolute value is continuously less than or equal to the predetermined threshold, and the continuous number of the data Is repeated every time when the number reaches a predetermined number,
The lancet counter detects the number of lancets, which is the number of times that the continuation number of the data is detected to reach the predetermined number by the run counter, corresponding to each of the plurality of predetermined threshold values. Is,
The minimum code length category determining means determines whether or not the number of lancets detected by the lancet counter corresponding to each of the plurality of types of predetermined thresholds is equal to or greater than a predetermined number. The minimum code length category is determined so that the category to which the category code of the minimum code length is assigned is a category representing a smaller number of significant bits as compared to when the lowest threshold value is high when it is low.
The data encoding apparatus according to claim 3 . 2. The information relating to the coding table created by the coding table creating means and the coded data generated by the coding means are output as one file. The data encoding device described in 1 .   A first step of detecting, by a run counter, a continuous number of the data whose data value is continuously within a predetermined range, and repeatedly performing the detection every time the continuous number of the data reaches a predetermined number;
  A second step of detecting, by the lancet counter, the number of lancets, which is the number of times that the continuation number of the data is detected by the lance counter to reach the predetermined number;
  According to the number of lancets detected by the lancet counter, a minimum code length category determining means determines a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data. And a third step
  A category and a category code are assigned so that a category code having a shorter code length is assigned to a category representing the number of significant bits close to the number of significant bits represented by the minimum code length category determined by the minimum code length category determining means. A fourth step of creating an encoding table indicating the correspondence relationship of
  A fifth step of generating encoded data by associating the category code with the significant bit portion of the data based on the encoding table generated by the encoding table generating means;
  Have
  A plurality of types of the predetermined range can be set,
  The run counter detects, for each of a plurality of set predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range, and the continuous number of the data is The detection is repeated every time a predetermined number is reached,
  The lancet counter detects the number of lancets corresponding to each of the predetermined ranges of the plurality of types, which is the number of times that the continuation number of the data is detected by the run counter to reach the predetermined number. And
  The minimum code length category determination means includes a minimum code length among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. Determines the minimum code length category that is the category to which the category code is assigned
  A data encoding method characterized by the above.   Computer
  A run counter for detecting the continuous number of the data whose value continuously falls within a predetermined range and repeatedly performing the detection every time the continuous number of the data reaches a predetermined number;
  A lancet counter for detecting the number of lancets, which is the number of times that the continuation number of the data is detected to have reached the predetermined number by the lance counter;
  A minimum code length category determining means for determining a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data according to the number of lancets detected by the lancet counter;
  A category and a category code are assigned so that a category code having a shorter code length is assigned to a category representing the number of significant bits close to the number of significant bits represented by the minimum code length category determined by the minimum code length category determining means. A coding table creating means for creating a coding table indicating the correspondence between
  Encoding means for generating encoded data by associating the category code with the significant bit portion of the data based on the encoding table created by the encoding table creating means;
  Is a program for functioning as
  A plurality of types of the predetermined range can be set,
  The run counter detects, for each of a plurality of set predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range, and the continuous number of the data is The detection is repeated every time a predetermined number is reached,
  The lancet counter detects the number of lancets corresponding to each of the predetermined ranges of the plurality of types, which is the number of times that the continuation number of the data is detected by the run counter to reach the predetermined number. And
  The minimum code length category determination means includes a minimum code length among categories representing the number of significant bits of the data according to the number of lancets detected corresponding to each of the plurality of types of predetermined ranges by the lancet counter. Determines the minimum code length category that is the category to which the category code is assigned
  A program characterized by that.   A run counter for detecting the continuous number of the data whose data value is continuously within a predetermined range, and repeatedly performing the detection every time the continuous data number reaches the predetermined number, and the run counter A lancet counter for detecting the number of lancets, which is the number of times that the continuous number of the data has reached the predetermined number, and a significant bit of the data according to the number of lancets detected by the lancet counter A minimum code length category determining means for determining a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing numbers, and a minimum code length category determined by the minimum code length category determining means. A category code with a shorter code length for a category representing a significant bit number close to the significant bit number A coding table creating means for creating a coding table showing a correspondence relationship between a category and a category code to be assigned, and the significance of the category code and the data based on the coding table created by the coding table creating means Encoding means for generating encoded data in association with the bit part, the predetermined range can be set to a plurality of types, and the run counter can be set to a plurality of set predetermined ranges. For each of the above, the number of continuous data whose data values are continuously within the predetermined range is detected, and the detection is repeated every time the number of continuous data reaches the predetermined number. The lancet counter is the number of lancets that is the number of times that the continuation number of the data is detected by the lance counter to reach the predetermined number. The minimum code length category determining means detects the lancet detected by the lancet counter corresponding to each of the plurality of types of predetermined ranges. The minimum code length category, which is the category to which the category code of the minimum code length is assigned among the categories representing the number of significant bits of the data, is generated by the data encoding device. A data decoding apparatus for decoding the encoded data, wherein the encoding table reproducing means reproduces the encoding table used for encoding based on the information related to the generated encoding table. When,
  Decoding means for decoding the encoded data and reproducing the data based on the encoding table reproduced by the encoding table reproducing means;
  A data decoding apparatus comprising:   A first step of detecting a continuous number of the data whose value continuously falls within a predetermined range by a run counter, and repeatedly detecting the data every time the continuous number of data reaches the predetermined number; A second step of detecting the number of lancets, which is the number of times that the continuation number of the data has reached the predetermined number by the ran counter, and a minimum code length category determining means, A third step of determining a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data according to the number of detected lancets; and the minimum code length category determining means The number of significant bits close to the number of significant bits represented by the minimum code length category determined by A fourth step of creating a coding table indicating a correspondence relationship between a category and a category code so that a category code having a shorter code length is assigned to a category to be represented; and a coding table created by the coding table creating means And a fifth step of generating encoded data by associating the category code with the significant bit portion of the data based on the category code, and the predetermined range can be set in a plurality of types. , For each of a plurality of set predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range is detected, and the continuous number of the data reaches the predetermined number The detection is repeated every time, and the lancet counter has reached the predetermined number of consecutive data by the run counter. Are detected corresponding to each of the plurality of types of predetermined ranges, and the minimum code length category determination means is configured to detect the plurality of types of predetermined lancets by the lancet counter. In accordance with the number of lancets detected corresponding to each of the ranges, a minimum code length category, which is a category to which a category code of the minimum code length is assigned among categories representing the number of significant bits of the data, is determined. A data decoding method for decoding encoded data generated by a data encoding method characterized by comprising:
  Based on the information related to the created encoding table, reproduce the encoding table used for encoding,
  A data decoding method characterized by decoding the encoded data and reproducing the data based on the reproduced encoding table.   A run counter for detecting a continuous number of the data whose data values are continuously within a predetermined range, and repeatedly performing the detection every time the continuous number of the data reaches the predetermined number, the run counter A lancet counter for detecting the number of lancets, which is the number of times that the number of consecutive data has reached the predetermined number, and a significant bit of the data according to the number of lancets detected by the lancet counter. A minimum code length category determining means for determining a minimum code length category which is a category to which a category code of the minimum code length is assigned among categories representing numbers, and the significance represented by the minimum code length category determined by the minimum code length category determining means A category representing a significant number of bits close to the number of bits has a shorter code length category. A coding table creating means for creating a coding table showing the correspondence between categories and category codes so as to assign a re-code, and the category code and the data based on the coding table created by the coding table creating means Is a program for functioning as an encoding means for generating encoded data by associating with significant bit portions of the predetermined bit range, and a plurality of types of the predetermined range can be set, and the run counter is set In addition, for each of a plurality of predetermined ranges, the continuous number of the data in which the data value is continuously within the predetermined range is detected, and each time the continuous number of the data reaches the predetermined number, The detection is repeated, and the lancet counter indicates that the number of consecutive data has reached the predetermined number by the lance counter. The number of detected lancets is detected in correspondence with each of the plurality of types of predetermined ranges, and the minimum code length category determination means uses the lancet counter to determine the plurality of types of predetermined ranges. In accordance with the number of lancets detected corresponding to each of the above, the minimum code length category, which is a category to which the category code of the minimum code length is assigned among the categories representing the number of significant bits of the data, is determined. A program for causing a computer to execute a process of decoding encoded data generated by a featured program,
  Based on the information related to the created encoding table, reproduce the encoding table used for encoding,
  A program for causing a computer to execute a process of decoding the encoded data and reproducing the data based on the reproduced encoding table.

Images