JP2012249061A - Run length coding device, run length decoding device, run length coding method, and run length decoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate run length code from raw image data efficiently.SOLUTION: A color code assignment unit 13 encodes raw image data with color code and run lengths to generate run length code, calculates a histogram of run lengths per run for each color code composing the run length code, and, on the basis of a calculated histogram, extracts a part of runs which have had some color code assigned thereto, thereby assigning unused color code to it. After code assignment by the color code assignment unit 13 is finished, an encoding unit 14 calculates the number of bits as a length part bit count for each color code, with which the coding result of the entire run will be at minimum, in which way it encodes each run.

Description

  The present invention is a technique related to run-length encoding.

  Run-length coding depends on the content of the image, but if it is a suitable image, it has features such as a high compression ratio, lossless compression, and a small circuit scale when implemented in hardware. Widely used. An outline of run-length encoding is described in Patent Documents 1 and 2, for example.

  Also in the camera, run-length encoding can be applied for the purpose of holding an OSD (on-screen display) display pattern to be superimposed and output on an image. If a predetermined OSD pattern is to be displayed, the OSD display pattern is run-length encoded in advance and stored as run-length data, and the run-length data is decoded in real time in accordance with the display timing. However, this is because the RAM for the frame buffer is unnecessary and the cost is low. That is, run-length encoding is suitable for such applications because it has features such as a small compression loss, a simple circuit configuration of a decoding circuit, and real-time processing.

  When the OSD display pattern is run-length encoded, since the pattern to be encoded is known in advance, the encoding parameters can be optimized from the characteristics of the entire image.

  The OSD display pattern displayed in color is often limited in the number of colors used, unlike a natural image. Therefore, when the OSD display pattern is run-length encoded, each color is represented using a color code, and the number of bits of the color code is set according to the number of colors used. Each color code is converted into color information using a color code table in which each color code is associated with color information for displaying each color code on a display, and run-length data is decoded into an OSD display pattern. The

  FIG. 29A shows an example of run-length encoding of an 8 × 8 OSD display pattern. In FIG. 29A, each grid represents a pixel, and the numbers in each grid indicate the color code of each pixel. FIG. 29B shows a run-length code obtained by encoding the OSD display pattern shown in FIG. 29A with a color code and a run-length.

  In this run length code, alphabets such as A and B indicate the color code of the run (A: color code 0, B: color code 1,...), And the subsequent numbers indicate the run length of the run in decimal. Show. For example, the first “B18” indicates a run with a color code of 1 and a run length of 18. Also, “A5” next to “B18” indicates a run with a color code of 0 and a run length of 5.

  In run-length encoding, a run may be divided for each horizontal line, and the next horizontal line may be a new run. However, in the example of FIG. 29, a run is not divided even when the horizontal line is changed, and is encoded as a continuous run. It has become. Even in this way, correct decoding is possible by designating the number of horizontal pixels at the time of decoding.

The run-length data in which the run-length code is represented by a bit pattern is represented by a bit pattern in which a predetermined number of bits are assigned to each of the color code part and the length part. The number of bits in the color code portion is determined by the number of colors used, and a value that can minimize the code length of the run-length data is adopted as the number of bits in the length portion. When the number of colors used is 8, the number of bits in the color code part is 3 bits. The relationship between the number of bits in the length part and the number of bits in the color code part when the number of bits in the length part is m and the run length of each run is 0 to 2 ^m−1 is as shown in FIG. Become.

  FIG. 30 shows the code length of run-length data for each color code when the number of bits in the color code part is 3 and the number of bits in the length part is changed for the run-length code in FIG.

  In FIG. 30, 1, 88, 168 are stored in the first row. This is because when m = 1 and the run-length code in FIG. 29 is represented by a bit pattern, the run-length data with a color code of 0 has a code length of 88 bits, and the run-length data with a color code of 1 has a code length of 168. It shows that it becomes a bit.

  In FIG. 30, 2, 40, 75 are stored in the second row. This is because when m = 1 and the run-length code in FIG. 29 is represented by run-length data, the run-length data with a color code of 0 has a code length of 40 bits, and the run-length data with a color code of 1 has a code length. Indicates 75 bits.

  As can be seen from the table of FIG. 30, for the color code 0, if the number of bits in the length portion is 3, the run length code of FIG. 29 can be represented by 24-bit run length data. If the number of bits in the length part is 5, it can be expressed by 64-bit run-length data, and the code length of the run-length data can be minimized.

  FIG. 31 shows run-length data of the run-length code of FIG. 29 when the number of bits of the length portion of the color code 0 is 3 and the number of bits of the length portion of the color code 1 is 5.

  As shown in FIG. 31, since the color code 1 represented by B has 5 bits in the length portion, B18 is represented by 8 bits of 00110010. Further, since the color code 0 represented by A has 3 bits in the length portion, A5 is represented by 6 bits of 00101.

  In this way, when the run-length codes “B18 to B17” in FIG. 29 are represented by run-length data, 64-bit run-length data shown in FIG. 31 is obtained.

JP 2010-131681 A JP 2009-060317 A

  When the number of bits in the color code part is 3, for example, an image of 8 colors can be handled. Therefore, when there are less than 8 colors used in the actual image, an unused color code is generated. In the conventional run-length encoding, how to handle this unused color code. It was not considered at all.

  In addition, the run length data of the OSD display pattern often has a bias in the run length distribution. FIG. 32A is a run histogram of color code 1 in the run-length code of FIG. FIG. 32B shows the code length when the number of bits of the color code part is 3 bits and the number of bits of the length part is 1 to 5 in the color code 1 of the run-length code of FIG.

  As can be seen from the table in FIG. 32B, the code length can be minimized by setting the number of bits in the length part to 5. However, a run with a run length of 2 to 3 is sufficient if the number of bits in the length portion is 2. Therefore, if the number of bits in the length portion is uniformly set to 5 for the color code 1, the number of bits in the length portion is extra 3 bits for runs with run lengths of 2 to 3, resulting in poor encoding efficiency. This is because the run-length distribution has a plurality of biases, but the plurality of biases were not considered at all in the conventional run-length encoding.

  An object of the present invention is to provide a technique capable of more efficiently run-length encoding original image data.

  (1) A run-length encoding apparatus according to the present invention is a run-length data representing each run by a color code part that stores a color code with a predetermined number of bits and a length part in which the number of bits is determined for each color code. A run-length encoding device that encodes original image data, encoding the original image data with a color code and a run length, generating a run-length code, and for each color code constituting the run-length code, A run length histogram of each run is calculated, a part of a run to which a certain color code is assigned is extracted based on the calculated histogram, and a color code assigning unit assigned to an unused color code, and the color code assignment After the allocation by the part is completed, for each color code, the number of bits that minimizes the encoding result of all runs is obtained as the number of bits of the length part. And a coding unit for coding the down.

  According to this configuration, the run of any one of the color codes is assigned to an unused color code based on the run length histogram for each color code. Therefore, the original image data can be efficiently encoded.

  (2) The encoding unit subtracts the base value of the corresponding color code from the run length for each run by subtracting the base value calculation unit that calculates the minimum value of the run length as the base value for each color code. A difference run length calculation unit for calculating a difference run length, and a difference in which each run is encoded by obtaining the minimum number of bits that can represent the difference run length of all runs as the number of bits of the length part for each color code It is preferable to include a run-length encoding unit.

  According to this configuration, the minimum run length is calculated as the base value for each color code. Then, a difference run length is calculated by subtracting the base value of the corresponding color code from the run length of each run. Therefore, the differential run length represents an offset from the base value, and is smaller than the run length. Therefore, by representing each run with a differential run length, the code length of each run can be further shortened, and the bit length of the run length data can be further shortened.

  (3) It is preferable that the encoding unit further includes a base value table generation unit that generates a base value table by associating the base value with the number of bits of the length part for each color code.

  According to this configuration, for each color code, a base value table in which the base value and the number of bits of the length part are associated with each other is generated. By referring to this base value table, the difference run length is changed to the run length. Can be calculated.

  (4) The color code assigning unit identifies a color code having a plurality of deviations in run length from the histogram, extracts a run belonging to the same deviation for the identified color code, and selects the unused color code. It is preferable to assign.

  According to this configuration, when there are a plurality of deviations in each color code, a run is assigned to each color code so that these deviations are distributed. Therefore, encoding efficiency can be improved.

  (5) The color code allocating unit calculates a code length reduction amount while changing a division position of the number of bits of the length portion for each color code, and selects a color code having a division position that maximizes the reduction amount. The color codes having a plurality of biases are extracted, and in the specified color code, a run having a run length represented by a bit number larger than a division position that maximizes the reduction amount is extracted, and the unused color is extracted. It is preferable to assign a code.

  According to this configuration, the division position for dividing the number of bits of the length portion in each color code can be obtained more accurately.

  (6) A run-length decoding apparatus according to the present invention uses a color code part that stores a color code with a predetermined number of bits and a length part in which the number of bits is determined for each color code to represent each run. A run-length decoding device that decodes length data into original image data, wherein the length section stores a differential run length obtained by subtracting the base value of each run from the run length of each run, and the base value is , A minimum value of run length in each color code, a base value table holding unit for holding a base value table in which the base value and the number of bits of the length part are associated with each color code, and each color code in bits A color information holding unit that holds color information to be expressed in a map format, a color code specified from the run-length data, and a bit of the length part corresponding to the specified color code Is determined from the base value table, and the run length data following the color code for the specified number of bits is extracted as a difference run length, and the difference run extracted by the difference run length extraction unit An adder that adds a base value corresponding to a length and calculates a run length; and a decoding that decodes the run length data into the original image data based on the run length calculated by the adder and the color information A part.

  According to this configuration, it is possible to decode the encoded run length in which each run is represented using the differential run length.

  According to the present invention, original image data can be efficiently encoded into run-length data.

It is a block diagram of the run-length encoding apparatus by embodiment of this invention. (A) is original image data in which color code 2 is assigned to a run having a color code of 1 and a run length of 2 to 3 with respect to FIG. (B) is a run-length code of the original image data of (A). This is the code length for each color code when the run-length code in FIG. 2B is encoded by changing the number of bits in the length part. It is the figure which showed the process by which a base value table is produced | generated. It is a block diagram of the run length decoding apparatus by embodiment of this invention. It is a functional block diagram of the vehicle-mounted camera of the comparative example to which the run length encoding apparatus shown in FIG. 5 is not applied. FIG. 6 is a functional block diagram of a vehicle-mounted camera to which the run-length decoding device shown in FIG. 5 is applied. It is a block diagram which shows the hardware constitutions of the vehicle-mounted camera to which the run length decoding apparatus shown in FIG. 5 was applied. FIG. 9 is a functional block diagram of an OSD display control unit 83 in a case where six types of OSD display patterns are superimposed and output on captured image data in the in-vehicle camera shown in FIG. 8. It is the block diagram which showed the process at the time of a run length encoding apparatus writing run length data in a vehicle-mounted camera. (A) is the flowchart which showed the run length encoding process which the run length encoding apparatus shown in FIG. 1 performs. 12 is a flowchart showing a subroutine of color code division processing in FIG. It is a flowchart which shows the subroutine of the base value calculation process shown in FIG. It is a flowchart which shows the subroutine of the run length encoding process of FIG. 17 is a flowchart showing a subroutine of the run extraction process shown in FIGS. 13, 14, and 16. 12 is a flowchart showing a subroutine of length part bit number calculation processing shown in FIG. It is the figure which showed the OSD display pattern used as the object of run length encoding in Example 2. FIG. It is an initial histogram. It is an initial color code table. It is an initial code length table. It is a code length reduction amount table. This is a histogram obtained by updating the initial histogram. 20 is a color code table obtained by updating the initial color code table shown in FIG. It is the histogram finally obtained by the color code division process. It is a color code table finally obtained by the color code dividing process. It is a code length table finally obtained by color code division processing. It is a table | surface which shows the effect at the time of calculating | requiring and encoding a difference run length with respect to the OSD display pattern shown in FIG. 17 in which the color code division | segmentation process was performed. 10 is a table summarizing the results of Example 2. (A) is an example of run-length encoding of an 8 × 8 OSD display pattern. (B) is a run length code obtained by encoding the OSD display pattern shown in FIG. 29 (A) with a color code and a run length. The run length data shown in FIG. 29 is the code length of run length data for each color code when the number of bits in the color code part is 3 and the number of bits in the length part is changed. 29 is a bit pattern of the run length code of FIG. 29 with the number of bits of the length portion of the color code 0 being 3 and the number of bits of the length portion of the color code 1 being 5. FIG. 29A is a histogram of the run of color code 1 in the run length code of FIG. (B) is the code length when the number of bits of the color code portion is 3 bits and the number of bits of the length portion is 1 to 5 in the color code 1 of the run length code of FIG.

  FIG. 1 is a block diagram of a run-length encoding apparatus according to an embodiment of the present invention. The run-length encoding apparatus includes an image acquisition unit 11, a table holding unit 12, a color code allocation unit 13, an encoding unit 14, and a run-length data holding unit 16.

  The image acquisition unit 11 acquires original image data to be encoded. Here, the original image data has, for example, a data structure in which a plurality of pixels are arranged in a matrix. Each pixel has a pixel value composed of four components of R, G, B, and transparency.

  The color code assigning unit 13 assigns a color code for each color to each pixel of the original image data acquired by the image acquiring unit 11. Then, the original image data with the color code is encoded with the color code and the run length to generate a run length code. Here, the run-length code is composed of a plurality of runs. Each run is composed of a color code and a run length. The color code indicates the color of the run. Run length indicates the length of the run.

  For example, in the example of FIG. 29A, when raster scanning is performed from the upper left pixel, 18 pixels of color code 1 are continuous. Therefore, these 18 pixels form one run. The code B corresponding to the color code 1 is set as the color code of this run. In the example of FIG. 29B, the color codes 0 and 1 are represented by the symbols A and B, respectively.

  Further, since this one run includes 18 pixels, the decimal number 18 is set as the run length of this one run. The color code assigning unit 13 repeats this process for the original image data to which the color code shown in FIG. 29A is assigned, and generates a run-length code shown in FIG.

  The color code assigning unit 13 calculates a run-length histogram of each run for each color code constituting the generated run-length code, and a part of the run to which a certain color code is assigned based on the calculated histogram. Are extracted and assigned to unused color codes.

  In the example of FIG. 29A, since the original image data includes color code 0 and color code 1, histograms of the respective color codes 0 and 1 are calculated. In this example, for the color code 1, for example, a histogram as shown in FIG. That is, the color code assigning unit 13 divides the run length bit length into a plurality of classes, identifies which class the run length of each run belongs to, and calculates a histogram by voting to the identified class, It is held by the holding unit 12.

  In the present embodiment, the color code is finally displayed as a binary number in a 3-bit bit string. Therefore, although the color code can represent a maximum of eight color codes, since only two color codes are used in the original image data of FIG. 29A, six color codes are unused. Yes.

  Therefore, when there is an unused color code, the color code allocating unit 13 assigns the run of one of the color codes to the unused color code based on the run length histogram for each color code, thereby encoding efficiency. Has improved.

  In the example of FIG. 32 (A), the run lengths are distributed in a biased way to 2-3 and 16-31. Therefore, the color code assigning unit 13 assigns an unused color code 2 to the color code of the run having a run length of 2 to 3. That is, the color code assigning unit 13 extracts runs belonging to the same bias and assigns unused color codes to color codes having a plurality of biases in the run length.

  FIG. 2A shows original image data when color code 2 is assigned to a run having a color code of 1 and a run length of 2 to 3 in the example of FIG. As shown in FIG. 2A, the pixel to which color code 1 is assigned in FIG. FIG. 2B is a run-length code of the original image data in FIG. In this example, a C code is assigned to the color code 2 to represent a run length code.

  In addition, the color code assigning unit 13 generates a color code table (see FIG. 19) in which each color code is associated with the number of bits of the length part, and stores the color code table in the table holding unit 12.

  Then, when the color code assignment is changed, the color code assignment unit 13 updates the color code table and the histogram so that the change in the assignment is reflected. Specifically, when the color code assignment for a certain run is changed, the color code assigning unit 13 updates the color code table so that the original color code of the run and the changed color code are associated with each other. .

  In this way, by updating the color code table, when decoding is performed, the color originally assigned to the run can be reproduced for the run whose color code assignment is changed.

  After the assignment by the color code assigning unit 13 is completed, the encoding unit 14 obtains, as the number of bits of the length part, the number of bits that minimizes the encoding result of all runs for each color code, encodes each run, Run-length data representing the run-length code as a bit pattern is obtained. Here, the run length data has a data structure in which each run is represented by a color code portion and a length portion. The number of bits of the color code part is 3.

  FIG. 3 shows the code length for each color code when the run-length code in FIG. 2B is encoded by changing the number of bits in the length portion. If color code 0 is set to 3 in the length part, color code 1 is set to 5 in the length part, and color code 2 is set to 2 in the length part, the run length data code length is 55. The code length can be minimized. On the other hand, before assigning a part of the run of color code 1 to color code 2, the code length of the run-length data was 64 bits as shown in FIG. Therefore, the coding efficiency can be improved by allocating the run having the run length of 2 to 3 to the color code 2 among the runs of the color code 1.

  Specifically, the encoding unit 14 includes a base value calculation unit 141, a differential run length calculation unit 142, a differential run length encoding unit 143, and a base value table generation unit 144.

  The base value calculation unit 141 generates a run length code from the original image data acquired by the image acquisition unit 11 in the same manner as the color code allocation unit 13, and calculates the minimum run length value for each color code as a base value. To do. The difference run length calculation unit 142 generates a run length code from the original image data acquired by the image acquisition unit 11 in the same manner as the color code allocation unit 13, and for each run, the base value of the corresponding color code from the run length. Is subtracted to calculate the differential run length of each run.

  The difference run length encoding unit 143 generates a run length code from the original image data acquired by the image acquisition unit 11 in the same manner as the color code allocation unit 13, and represents the difference run length of all runs for each color code. The minimum number of bits that can be obtained is obtained as the number of bits in the length part, and each run is encoded.

  The base value table generation unit 144 associates the base value calculated for each color code by the base value calculation unit 141 with the number of bits of the length part calculated for each color code by the differential run length encoding unit 143. A table is generated and held in the table holding unit 12.

  FIG. 4 is a diagram illustrating processing for generating a base value table. 4A is the original image data to which the color code of FIG. 2A is attached, and FIG. 4B is a run-length code when the original image data of (A) is expressed using a differential run-length. Yes, (C) is a base value table.

  In the example of FIG. 4C, the base values of the color codes 0, 1, and 2 are calculated as 5, 17, and 2, respectively. Therefore, the run difference run length of “B18” in FIG. 2B is 18−17 = 1. Therefore, this run of B18 is represented as B1 in FIG. Differential run lengths are similarly obtained for other runs. For each color code, the number of bits that can represent the total differential run length is obtained. Then, as shown in FIG. 4C, the number of bits of the length portion of the color codes 0, 1, and 2 is calculated as 1, respectively. As described above, as shown in FIG. 4C, a base value table in which the number of bits of the length part and the base value are associated with each color code is generated.

  When the run-length code shown in FIG. 4B is represented by run-length data, the code length is 36 bits. Therefore, the code length reduced to 55 bits by distributing the color code in FIG. 2B is further reduced to 36 bits, and the encoding efficiency is improved.

  Returning to FIG. 1, the run-length data holding unit 16 holds the run-length data generated by the differential run-length encoding unit 143.

  In this way, by representing each run using the differential run length representing the run length offset from the base value, the code length of each run can be further shortened, and the code length of the run length data can be further shortened. .

  FIG. 5 is a block diagram of a run-length decoding apparatus according to the embodiment of the present invention. The run-length decoding apparatus decodes run-length data generated by the run-length encoding apparatus shown in FIG.

  Specifically, the run length decoding apparatus includes a base value table holding unit 51, a color information holding unit 52, a difference run length extracting unit 53, an adding unit 54, and a decoding unit 55.

  The base value table holding unit 51 holds a base value table in which the base value and the number of bits of the length part are associated with each color code. That is, the base value table holding unit 51 holds the base value table generated by the run-length encoding device shown in FIG.

  As shown in FIG. 5, the base value table holding unit 51 stores the same base value table as that shown in FIG.

  The color information holding unit 52 holds color information necessary for developing run-length data into bitmap data. As shown in FIG. 5, the color information holding unit 52 holds a color information table in which four pixel values of R, G, B pixel values and pixel values indicating transparency α are associated with each color code. ing.

  The difference run-length extraction unit 53 identifies the color code from the run-length data generated by the run-length encoding apparatus, identifies the number of bits of the length part corresponding to the identified color code from the base value table, and identifies Run-length data following the color code for the number of bits is extracted as a differential run-length.

  Specifically, the differential run length extraction unit 53 first extracts the code string stored in the first 3 bits of the run length data. In the present embodiment, each run is represented in the order of a color code part and a length part in the run length data, and the number of bits of the color code part is 3 bits. Therefore, the first 3 bits of the run length data represent the color code. In the example of FIG. 5, a code string of 001 is stored in the first 3 bits. Therefore, the differential run length extraction unit 53 determines that the color code is 1 for the first run.

  Next, the differential run length extraction unit 53 specifies the number of bits of the length portion corresponding to the color code 1 from the base value table. In this case, the number of bits of the length portion corresponding to the color code 1 is 1. Therefore, the differential run length extraction unit 53 determines that the number of bits in the length portion of the first run is 1.

  Next, the differential run length extracting unit 53 extracts 1-bit run length data following the color code of the first run as a differential run length. In this case, the differential run length is 1. That is, the first run is represented by a code string having a color code of 001 and a length part having a code string of 1. Therefore, the differential run length extracting unit 53 determines the fifth, sixth, and seventh third bits of the run length data as the color code portion of the second run. The difference run length extraction unit 53 repeats the above processing, and sequentially specifies the color code and the difference run length included in the run length data.

  The adding unit 54 adds the base values corresponding to the differential run length extracted by the differential run length extracting unit 53, and calculates the run length. In the example of FIG. 5, the differential run length of the first run is 1. The color code of the first run is 1. Therefore, the adding unit 54 reads 17 which is the base value of the length portion corresponding to the color code 1 from the base value table. Then, the adding unit 54 adds the difference run length 1 to the read base value 17 to calculate the run length. In this case, the run length is 18, and “B18” shown in FIG. 2B is reproduced.

  The decoding unit 55 expands the run length data into bitmap data based on the run length and color information table calculated by the addition unit 54. In the previous example, the color code for the first run was 1 and the length was 18.

  Therefore, the decoding unit 55 reads out the R, G, B, and α pixel values corresponding to the color code 1 from the color information table, and R and G corresponding to the color code 1 for 18 consecutive pixels. , B, α pixel values are set, and each pixel is output sequentially, whereby the run-length data is developed into bitmap data. Thereby, original image data is obtained.

  For example, the decoding unit 55 develops bitmap data by writing pixel values into a frame buffer in which captured image data captured by the imaging system is expanded. As a result, a dedicated frame buffer for the OSD display pattern becomes unnecessary. And the decoding part 55 repeats said process with respect to each run, and superimposes and displays an OSD display pattern on captured image data.

Example 1
Next, a first embodiment in which the run-length decoding apparatus shown in FIG. 5 is applied to a vehicle-mounted camera will be described. A vehicle-mounted camera is required to have a function of superimposing a graphic on a captured image and outputting it, such as a vehicle width line display for backview monitor applications. This graphic is an OSD display pattern.

  As a method for displaying the OSD display pattern, a configuration in which a frame buffer dedicated to the OSD display pattern is provided in the ECU or the like and the OSD pattern is drawn in the frame buffer by a graphic controller is adopted.

  FIG. 6 is a functional block diagram of a vehicle-mounted camera of a comparative example to which the run-length encoding device shown in FIG. 5 is not applied. The in-vehicle camera shown in FIG. 6 includes an imaging system 61, an image processing circuit 62, a graphic controller 63, a frame buffer 64, and an output control unit 65.

  The imaging system 61 is configured by a camera that images a subject at a predetermined frame rate, and acquires captured image data. The image processing circuit 62 performs predetermined image processing such as gamma correction on the captured image data acquired by the imaging system 61. The graphic controller 63 includes a drawing circuit 631 and a superimposing circuit 632.

  The drawing circuit 631 receives a drawing command from the CPU and generates an OSD display pattern in the frame buffer 64. The superimposing circuit 632 superimposes the OSD display pattern developed in the frame buffer 64 and the captured image data output from the image processing circuit 62, and outputs them to the output control unit 65. Specifically, the superimposing circuit 632 reads an OSD display pattern from the frame buffer 64 in synchronization with the captured image data input from the image processing circuit 62, and generates a superimposition result by pixel calculation. The output control unit 65 outputs the superimposed result image to the outside as an NTSC signal or a digital signal.

  Thus, the configuration of the in-vehicle camera of the comparative example requires a large-capacity RAM as the frame buffer 64. In addition, the graphic controller 63 including the drawing circuit 631 is required, which increases the cost.

  The OSD display pattern required for in-vehicle use employs a predetermined graphic image such as a vehicle width line. Therefore, if a configuration is adopted in which a necessary OSD display pattern is encoded and stored as data, and decoded and displayed in real time at the time of display, the frame buffer 64 and the drawing circuit 631 are not necessary, and the cost can be reduced.

  FIG. 7 is a functional block diagram of a vehicle-mounted camera to which the run-length decoding device shown in FIG. 5 is applied. The in-vehicle camera shown in FIG. 7 includes an imaging system 71, an image processing circuit 72, a superimposing circuit 73, a code data RAM 74, a decoding circuit 75, and an output control unit 76.

  The imaging system 71, the image processing circuit 72, and the output control unit 76 are the same as those in FIG. The code data RAM 74 holds run length data generated in advance by the run length encoding apparatus shown in FIG. The decoding circuit 75 is composed of a run-length decoding device shown in FIG. The superimposing circuit 73 superimposes the captured image data output from the image processing circuit 72 and the OSD display pattern output from the decoding circuit 75, and outputs a superimposition result.

  According to this configuration, the decoding of the OSD display pattern is performed in real time at the time of display inside the vehicle-mounted camera. This eliminates the need for the drawing circuit 631 and the frame buffer 64 dedicated to the OSD display pattern as in the in-vehicle camera of the comparative example, thereby simplifying the device configuration and reducing the cost.

  In addition, since the run-length encoded OSD display pattern held in the code data RAM 74 is generated outside the vehicle-mounted camera, even if the encoding algorithm is complicated, there is little influence on the cost. Moreover, since decoding of run-length data can be realized by a relatively simple process, real-time decoding is possible at the time of display, which is suitable for a vehicle-mounted camera.

  FIG. 8 is a block diagram showing a hardware configuration of a vehicle-mounted camera to which the run-length decoding device shown in FIG. 5 is applied.

  The in-vehicle camera shown in FIG. 8 includes an imaging system 81, an image processing circuit 82, an OSD display control unit 83, an output control unit 84, a CPU 85, a RAM 86, a ROM 87, and an I / O 88. The imaging systems 81 to I / O 88 are connected to each other via a bus line 89.

  The imaging system 81 and the image processing circuit 82 constitute the imaging system 71 and the image processing circuit 72 of FIG. 7, respectively. The OSD display control unit 83 constitutes the decoding circuit 75, the code data RAM 74, and the superimposing circuit 73 shown in FIG. The output control unit 84 constitutes the output control unit 76 shown in FIG. The CPU 85 governs overall control of the in-vehicle camera. The RAM 86 is used as a working memory for the CPU 85. The ROM 87 holds run length data created in advance. The I / O 88 inputs and outputs various control signals with the ECU, for example. The output control unit 84 outputs image data obtained by superimposing the OSD display pattern and the captured image data to the outside as a digital image signal, or converts the image data to an NTSC signal and outputs it to the outside.

  FIG. 9 is a functional block diagram of the OSD display control unit 83 when the six types of OSD display patterns are superimposed on the captured image data and output in the in-vehicle camera shown in FIG.

  The OSD display control unit shown in FIG. 9 decodes and outputs six types of OSD display patterns in real time during display. The OSD display control unit is provided with six stages of decoding units including a code data RAM 74, a decoding circuit 75, and a superimposing circuit 73 shown in FIG.

  An imaging system 71 shown in FIG. 7 is connected to the first stage superimposing circuit 73 via an image processing circuit 72, and captured image data captured by the imaging system 71 is input.

  The code data RAM 74 in the first to sixth stages respectively stores run-length data obtained in advance by run-length encoding the first to sixth OSD display patterns, which are six types of OSD display patterns. keeping. When receiving the decoding start control signal from the CPU 85, the first to sixth stage code data RAM 74 outputs the run-length data to be held to the decoding circuit 75, and the decoding circuit 75 performs the decoding process. Is executed.

  The decoding circuit 75 decodes the run length data and outputs it to the superimposing circuit 73. The first stage superimposing circuit 73 superimposes the first OSD display pattern output from the first stage decoding circuit 75 and the captured image data, and outputs the result to the second stage superimposing circuit 73. The second to fifth superimposing circuits 73 superimpose the second to fifth OSD display patterns on the image data output from the upper superimposing circuit 73 and output the image data to the lower superimposing circuit 73.

  The sixth stage superimposing circuit 73 superimposes the sixth OSD display pattern on the image data output from the fifth stage superimposing circuit 73 and outputs the image data to the output control unit 76. As a result, the image data in which the first to sixth OSD display patterns are superimposed on the captured image data is output to the output control unit 76.

  FIG. 10 is a block diagram showing processing when the run-length encoding apparatus writes run-length data to the vehicle-mounted camera.

  The computer 101 implements the function of the run length encoding device shown in FIG. First, the computer 101 activates a graphic creation tool and generates an OSD display pattern in accordance with an operation instruction from a designer (step 1).

  Next, the computer 101 performs run-length encoding on the generated OSD display pattern to generate run-length data (step 2). Next, the computer 101 writes the generated run length data in the ROM of the in-vehicle camera 102 (step 3). Thereby, the vehicle-mounted camera 102 obtains run length data. At this time, the computer 101 also writes a color code table and a base value table necessary for decoding the run-length data in the ROM 87.

  The computer 101 is connected to, for example, an I / O 88 included in the vehicle-mounted camera 102 via a cable, and outputs run-length data or the like to the vehicle-mounted camera 102.

  Next, the OSD display control unit 83 reads run-length encoded data from the ROM 87 (step 4), decodes it to generate an OSD display pattern, and outputs it to the outside via the output control unit 84.

(Example 2)
Next, a second embodiment in which the OSD display pattern used in the vehicle-mounted camera of the first embodiment is run-length encoded using the run-length encoding device shown in FIG.

  The specifications of the in-vehicle camera 102 shown in FIG. 10 are as follows.

Number of superimposed OSD display patterns: 6 Number of simultaneously displayable colors: 8 colors / surface Number of color code bits: 3 bits
Code data RAM 74: Since the color code is 3 bits, a RAM of 3 bits / word is used. In order to simplify the circuit, the length of each run is a multiple of 3 bits.

Code data RAM 74 size: 2049 bits / plane (3 bits / word × 683 words)
Display color: RGB 24-bit → Color information table corresponding to each color code on each surface is provided, and a 3-bit color code is converted into RGB 24-bit pixel data.

  Transmission display control: 0 (transmittance 0) to 1 (transmittance 100%) is designated in 17 steps of 1/16 steps.

Output image data: 360 × 480 pixels, 24 bits / pixel
Thus, since the size of the code data RAM 74 is limited to 2 KB, each OSD display pattern must be encoded to a size of 2 KB or less, and an encoding algorithm with high encoding efficiency is required.

<Run-length encoding>
FIG. 11A is a flowchart showing a run-length encoding process performed by the run-length encoding apparatus shown in FIG. FIG. 17 is a diagram illustrating an OSD display pattern that is a target of run-length encoding in the second embodiment.

  This OSD display pattern is composed of 360 × 480 pixels and has three color codes. A color code 5 is given to the graphic composed of the inclined wavy lines shown in the center of the screen. A color code 6 is assigned to the graphic of the outer frame covering the outer periphery of the OSD display pattern. A color code 0 is assigned to the graphic of the background of the graphic composed of the inclined wavy lines of the OSD display pattern.

  First, the color code assigning unit 13 performs color code division processing (S1). Next, the encoding unit 14 performs base value calculation processing (S2). Next, the encoding unit 14 performs run-length encoding to generate run-length data (S3).

<Color code division processing>
FIG. 12 is a flowchart showing a subroutine of color code division processing in FIG. First, the color code assigning unit 13 performs initialization. At this time, the color code assigning unit 13 generates an initial color code table (S11).

  FIG. 19 is an initial color code table. The color code table includes a color code field and a length bit number field. The color code table field stores eight color codes of 0 (A) to 7 (H). In the length part bit number field, a color code assigned to each length part bit number is shown in each color code.

  In the example of FIG. 19, the number of bits in the length part is divided every 3 bits, such as 0 bits, 3 bits, and 6 bits. This is because the code data RAM is set to 3 bits / word.

  Since FIG. 19 is an initial color code table, the color codes 0 to 7 are assigned to the color codes 0 to 7 regardless of the number of bits in the length part.

  In S11, the color code assigning unit 13 generates a code length table shown in FIG. FIG. 20 is an initial code length table. The code length table has the same data structure as the color code table. This code length table shows the code length of run-length data when the number of bits in the length portion is changed for each color code.

  For example, dda90 is stored in the cell in the first row and the first column. This is because the code length of the run length data of the color code 0 graphic is dda90 bits when the number of bits of the length portion is set to 0 bit for the color code 0.

  As shown in FIG. 20, for the color code 0, when the number of bits in the length part is 9 bits, the code length is 1ef0, which is the minimum. Therefore, the color code assigning unit 13 sets the number of bits of the length part of the color code 0 to 9.

  Similarly, for the color codes 5 and 6, when the number of bits in the length portion is set to 6 bits, the code length is minimized. Therefore, the color code assigning unit 13 sets the number of bits of the length part of the color codes 5 and 6 to 6, respectively.

  For each of the color codes 0, 5, and 6, the total code length of the run-length data when the number of bits in the length part is set to 9, 6, and 6 is 1ef0 + 06e4 + 14c7 = 449b = 17,563 bits.

  Returning to FIG. 12, the color code assigning unit 13 creates a histogram (S12). FIG. 18 is an initial histogram. The histogram has a data structure similar to that of the color code table. The numerical value stored in each cell of the histogram is a hexadecimal number.

  In the OSD display pattern of FIG. 17, in the color code 0 graphic, the number of runs that can be represented by the number of bits of the length portion greater than 0 bits and 3 bits or less is 6, so that the first row and second column 6 is stored in the cell. Also, in the graphic of color code 0, the number of bits that can be represented by the number of bits of the length portion greater than 3 bits and 6 bits or less was 48, so 48 is stored in the cell in the first row and the fourth column. Has been.

  In addition, since the OSD display pattern shown in FIG. 17 includes a color code 5 graphic and a color code 6 graphic in addition to the color code 0 graphic, numerical values are stored in the corresponding cells of the color code 5 and 6. Has been. Numeric values are not stored in cells of color codes other than color codes 0, 5, and 6.

  Returning to FIG. 12, the color code assigning unit 13 determines whether there is an unused color code (S13). When there is an unused color code (YES in S13), the color code assignment unit 13 calculates a color code division position (S14). In the initial histogram shown in FIG. 18, since no numerical value is stored in cells of color codes other than color codes 0, 5, and 6, the color code allocation unit 13 determines that an unused color code exists. .

  On the other hand, if there is no unused color code (NO in S13), the process is returned. In this case, if a numerical value is stored in any cell of the color code to which no color code is assigned in the initial state in the histogram, it is determined that there is no unused color code.

<S14> Color Code Division Position Calculation Processing Hereinafter, the color code division position calculation processing in S14 will be described in detail. First, the color code allocating unit 13 shows how much the code length can be reduced with respect to the number of bits of the current length part while changing the division position of the number of bits of the length part for each color code. A code length reduction amount table is calculated. Then, based on the code length reduction amount table, a color code division position capable of maximizing the reduction amount is calculated. FIG. 21 is a code length reduction amount table. The code length reduction amount table has the same data structure as the color code table.

  The color code assigning unit 13 currently sets the number of bits of the length part to 9, 6, 6 for the color codes 0, 5, 6. Therefore, the code length reduction amount table shown in FIG. 21 indicates the code length reduction amount when compared with this case.

  In color code 0, if the division position is 3 bits, the reduction amount is 36 bits. Here, the division position is 3 bits. In the color code 0, a run whose length part has 3 bits or less is run-length encoded with the length part having 3 bits, and the length part has 3 bits. A run with the number of bits larger than 3 bits indicates a case where run-length encoding is performed with the number of bits in the length part before division (9 bits in the case of color code 0).

  Similarly, in color code 0, if the division position is 6 bits, the reduction amount is 234 bits. Therefore, for the color code 0, the reduction amount can be maximized if the division position is 6 bits. For the color codes 5 and 6, the reduction amount is obtained while changing the number of bits in the length portion as in the case of the color code 0. As a result, the reduction amount can be maximized by setting the division position for the color code 5 to 3 bits and the division position for the color code 6 to 6 bits or 9 bits.

  Then, the color code assigning unit 13 determines the color code having the largest reduction amount as the color code to be divided among all the color codes in the code length reduction amount table shown in FIG. In this case, since the 6 or 9-bit reduction amount of the color code 6 is 1122 bits, the color code 6 is determined as the color code to be divided.

  Then, the color code assigning unit 13 determines 6 bits or 9 bits with the maximum reduction amount as the division position in the color code 6. In the color code 6, the reduction amount is the same even if the division position is 6 bits and 9 bits. In such a case, the color code allocating unit 13 uses the smaller 6 bits as the dividing position, but is not limited thereto, and the larger 9 bits may be used as the dividing position.

<S15> Histogram Division Processing Returning to FIG. 12, in S15, the color code allocation unit 13 sets the division position of the color code 6 to 6 bits, and therefore extracts a run having a run length larger than 6 bits in the color code 6. Assign to an unused color code. In this case, for example, the color code assigning unit 13 may assign the extracted run to the color code with the smallest number among the unused color codes. In this case, since the color code with the smallest number among the unused color codes is the color code 1, in the color code 6, a run having a run length larger than 6 bits is assigned the color code 1.

  Next, the color code assigning unit 13 updates the initial histogram shown in FIG. 18 to reflect that a new color code has been assigned to the extracted run. FIG. 22 is a histogram obtained by updating the initial histogram.

  As shown in FIG. 22, the numerical value stored in each cell having a length portion of 9-21 bits in the color code 6 is assigned to each cell having a length portion of 9-21 bits in the color code 1. The numeric value stored in each cell in which the number of bits in the length portion is 9 to 12 bits in the color code 6 is updated to 0.

<S16> Color Code Table Update Processing Returning to FIG. 12, in S16, the color code assignment unit 13 updates the initial color code table shown in FIG. FIG. 23 is a color code table obtained by updating the initial color code table shown in FIG.

  As shown in FIG. 23, in the color code 6, the numerical value stored in each cell having the length portion of 9 to 12 bits is updated from 6 to 1. As a result, even if color code 1 is assigned to a run having a run length larger than 6 bits in color code 6, it can be seen that these runs were originally color code 6.

  Next, the color code allocation unit 13 generates a code length table for each color code in the same manner as when the initial code length table shown in FIG. 20 is generated, and the length that minimizes the code length for each color code. The number of bits in the length part is obtained and set as the current number of bits in the length part of each color code.

  Thereafter, the color code assigning unit 13 repeats the processes of S13 to S16 while an unused color code remains. Finally, the histogram shown in FIG. 24 and the color code table shown in FIG. 25 were obtained. FIG. 24 is a histogram finally obtained by the color code dividing process. FIG. 25 is a color code table finally obtained by the color code dividing process.

  In the histogram shown in FIG. 24, it can be seen that each of the color code 1 to the color code 7 stores a numerical value in one of the cells, and is assigned a run. In addition, in the color code 0, all the cells are 0 because the value of the 3-bit cell of the color code 0 was moved to the unused color code 7 in the last loop of S13 to S16. is there.

  In the histogram shown in FIG. 24, it can be seen that runs having similar run lengths are assigned to the color codes 1 to 7. Therefore, the length portion of each color code can store the run length without waste.

  As shown in FIG. 25, for example, the color code 0 is stored in a 3-bit cell. Therefore, in the run-length data, it can be seen that the original color code is 0 in a run in which the color code is 7 and the number of bits of the length part is 3 bits.

  FIG. 26 is a code length table finally obtained by the color code dividing process. For each of the color codes 0 to 7, the code length can be minimized by setting the number of bits of the length portion by the number of bits corresponding to the shaded cells.

  The code length of the OSD display pattern shown in FIG. 17 is 001e + 1b48 + 0426 + 0288 + 01d4 + 1047 + 0024 = 3453 = 13,395 bits after color code division processing.

<Base value processing>
Next, the base value process will be described. FIG. 11B is a flowchart of the base value process. First, the base value calculation unit 141 performs a base value calculation process (S21). Next, the base value calculation unit 141 performs a length part bit number calculation process (S22).

<Base value calculation processing>
FIG. 13 is a flowchart showing a subroutine of the base value calculation process shown in FIG. First, the base value calculation unit 141 initializes various parameters (S31).

  Next, the base value calculation unit 141 determines whether or not the extraction of the last run of the OSD display pattern to be processed has ended, and if it has ended (YES in S32), returns the processing. On the other hand, when the extraction of the last run has not ended (NO in S32), the base value calculation unit 141 performs a run extraction process (S33).

<Run extraction process>
FIG. 15 is a flowchart showing a subroutine of the run extraction process shown in FIGS. 13, 14, and 16. First, 0 is assigned to the run length variable and initialized, and the pixel value of the target pixel is assigned to the color code variable to initialize the color code variable (S51). Here, the target pixel is determined in the raster scan order in the OSD display pattern.

  Next, 1 is added to the run length variable, and the run length variable is updated (S52). Next, the pixel position of the target pixel is updated (S53). Next, if the pixel value of the target pixel is the same value as the color code variable (YES in S54), the process returns to S52. On the other hand, if the pixel value of the target pixel is different from the color code variable, it is determined that one run has been extracted, and the process is returned. That is, when the target pixel is sequentially shifted with respect to the OSD display pattern and the pixel value of the target pixel becomes different from the pixel value of the previous target pixel, it is determined that one run has been extracted. Each time the pixel of interest is shifted, the value of the run length variable added by 1 becomes the run length of the extracted run.

  Returning to FIG. 13, in S34, the base value calculation unit 141 refers to the color code table shown in FIG. 25, and calculates the color code after the color code division processing of the run extracted in S33. For example, when a run that can be represented by a color code of 0 and a run length of greater than 3 bits and less than 6 bits is extracted, the color code of this run is calculated as 4.

  Next, when the run length of the run extracted in S33 is less than the current base value of the corresponding color code (YES in S35), the base value calculation unit 141 uses the run length of the run from which the current base value is extracted. Update (S36), and return the process to S32.

  On the other hand, if the run length of the run extracted in S33 is greater than or equal to the current base value of the corresponding color code (NO in S35), the current base value is not updated, and the process returns to S32.

  For example, if the run length of the run calculated as color code 7 in S34 is less than the current base value of color code 7, it is determined YES in S35, and the current base value of color code 7 is updated with this run length. Is done. Such processing is performed on each color code, and the minimum run length is calculated as the base value for each color code.

<Length part bit number calculation processing>
Next, the length part bit number calculation processing in S22 of FIG. 11B will be described. FIG. 16 is a flowchart showing a subroutine of length part bit number calculation processing shown in FIG. First, the differential run length calculation unit 142 initializes various parameters (S61). Next, the differential run length calculation unit 142 determines whether or not the extraction of the last run of the OSD display pattern to be processed has ended, and if it has ended (YES in S62), the process proceeds to S68. On the other hand, if the last run has not been extracted (NO in S62), the differential run length calculation unit 142 performs a run extraction process (S63). This run extraction process is the same as the process shown in FIG.

  Next, the difference run length calculation unit 142 calculates the color code of the run extracted in S63 (S64). This process is the same as S34 in FIG.

  Next, the difference run length calculation unit 142 calculates a difference run length by subtracting the base value of the corresponding color code from the run length extracted in S63 (S65). For example, in the run extracted in S63, when the color code calculated in S64 is the color code 7, the difference run length is calculated using the base value of the color code 7.

  Next, if the difference run length obtained in S65 is larger than the maximum difference run length of the corresponding color code (YES in S66), the difference run length encoding unit 143 sets the maximum difference run length of the corresponding color code to S65. (S67) and the process returns to S62.

  On the other hand, if the difference run length obtained in S65 is less than or equal to the maximum difference run length of the corresponding color code (NO in S66), the process returns to S62 without updating the maximum difference run length.

  In S68, the difference run length encoding unit 143 calculates, for each color code, the minimum number of bits that can represent the maximum difference run length as the number of bits of the length part, and returns the process.

<Run-length encoding process>
Next, details of the run length encoding process of S3 of FIG. 11 will be described. FIG. 14 is a flowchart showing a subroutine of the run-length encoding process of FIG. First, the differential run length encoding unit 143 initializes various parameters (S41). Next, the differential run length encoding unit 143 determines whether or not the extraction of the last run of the OSD display pattern to be processed has ended, and if it has ended (YES in S42), returns the processing. On the other hand, if the last run has not been extracted (NO in S42), the differential run length encoding unit 143 performs a run extraction process (S43). This run extraction process is the same as the process shown in FIG.

  Next, the differential run length encoding unit 143 calculates the color code of the run extracted in S43 (S44). This process is the same as S34 in FIG.

  Next, the differential run length encoding unit 143 outputs the color code calculated in S44 to the run length data holding unit 16 (S45). Next, the differential run length encoding unit 143 obtains a subtracted differential run length from the base value of the corresponding color code from the run length of the run extracted in S43. Then, by encoding the difference run length with the number of bits of the corresponding color code length portion, the difference run length is encoded and the length portion is calculated (S46). Next, the differential run length encoding unit 143 outputs the length portion calculated in S46 to the run length data holding unit 16 (S47).

  In this manner, the runs are sequentially extracted from the OSD display pattern, and the color code and the length portion of the extracted runs are sequentially output to calculate run length data.

  FIG. 27 is a table showing the effect when the difference run length is obtained and encoded with respect to the OSD display pattern shown in FIG. 17 subjected to the color code division processing.

  In FIG. 27, “number of bits in the length part (before change)” indicates the number of bits in the length part before the difference run length is obtained. “Run length” indicates a range of run lengths before the differential run length is obtained. The “base value” is the minimum value of the run length used when obtaining the differential run length. “Maximum differential run length” is the maximum value of the differential run length. “Number of bits in the length part (after change)” indicates the number of bits in the length part when the differential run length is obtained. “Run number” indicates the number of runs in each color code. “Code length reduction amount” indicates the reduction amount of the code length by adopting the differential run length for each color code.

  For example, when the run length is encoded without obtaining the differential run length for the color code 1, the number of bits of the length portion is 12 bits. On the other hand, when the color code 1 is encoded by obtaining the difference run length, the number of bits in the length part becomes 0 bits. Since the number of runs of color code 1 is 2, the code length reduction amount is 2 × 12 = 24 bits.

  Similarly, for color code 6, the code length reduction amount is 463 × 6 = 2778 bits. For the other color codes, the code length reduction amount was zero. Therefore, the total code length reduction amount was 2802.

  On the other hand, the code length of the run-length data when run-length encoding is performed without obtaining the differential run-length is 13395 bits. Therefore, by obtaining the differential run length, the code length of the run length data is 13395-2802 = 10,593 bits.

  FIG. 28 is a table summarizing the results of Example 2. When color code division processing and base value processing are not applied, that is, when conventional run-length encoding is used, the code length of the OSD display pattern shown in FIG. 17 is 17,563 bits. On the other hand, when only the color code division processing is applied, the code length of the OSD display pattern shown in FIG. 17 is 13,395 bits. When the color code division process and the base value process are applied, the code length of the OSD display pattern shown in FIG. 17 is 10,593 bits.

  In this embodiment, the color code division processing, the base value processing, and the run length encoding processing are combined. However, the present invention is not limited to this, and only the color code division processing is performed to generate run length data. Also good. Alternatively, only the base value processing and the run length encoding processing may be performed to generate run length data.

  In the above description, the minimum number of bits that can represent the difference run length is adopted as the number of bits of the length portion for each color code. Alternatively, the number of bits that minimizes the encoding result of all runs may be adopted as the number of bits in the length portion.

DESCRIPTION OF SYMBOLS 11 Image acquisition part 12 Table holding part 13 Color code allocation part 14 Encoding part 16 Run length data holding part 51 Base value table holding part 52 Color information holding part 53 Differential run length extraction part 54 Addition part 55 Decoding part

Claims (8)

A run-length encoding device that encodes original image data into run-length data representing each run with a color code part that stores a color code with a predetermined number of bits and a length part in which the number of bits is determined for each color code Because
The original image data is encoded with a color code and a run length, a run length code is generated, a run length histogram of each run is calculated for each color code constituting the run length code, and based on the calculated histogram A color code assigning unit that extracts a part of a run to which a certain color code is assigned and assigns it to an unused color code;
After the assignment by the color code assigning unit is finished, for each color code, obtain the number of bits that minimizes the encoding result of all runs as the number of bits of the length part, and an encoding unit that encodes each run. A run-length encoding device. The encoding unit includes:
For each color code, a base value calculation unit that calculates a minimum run length as a base value,
For each run, a difference run length calculation unit that subtracts the base value of the corresponding color code from the run length and calculates the difference run length of each run;
The difference run length encoding part which calculates | requires the minimum bit number which can represent the difference run length of all the runs for each color code as a bit number of the said length part, and encodes each run. Run-length encoding device.   The run-length encoding apparatus according to claim 2, wherein the encoding unit further includes a base value table generating unit that generates a base value table by associating the base value with the number of bits of the length part for each color code. .   The color code assigning unit identifies color codes having a plurality of deviations in run length from the histogram, extracts runs belonging to the same deviation for the specified color codes, and assigns the unused color codes. The run-length encoding apparatus according to any one of 1 to 3.   The color code allocating unit calculates a code length reduction amount while changing the division position of the number of bits of the length portion for each color code, and assigns the color code having a division position that maximizes the reduction amount to the plurality of color codes. In the specified color code, a run having a run length represented by a bit number larger than the division position that maximizes the reduction amount is extracted, and the unused color code is assigned. The run-length encoding apparatus according to claim 4. Run-length decoding that decodes run-length data representing each run into original image data using a color code part that stores a color code with a predetermined number of bits and a length part in which the number of bits is determined for each color code Device.
The length part stores a differential run length obtained by subtracting the base value of each run from the run length of each run,
The base value indicates the minimum run length of each color code,
For each color code, a base value table holding unit that holds a base value table in which the base value and the number of bits of the length part are associated;
A color information holding unit that holds color information for representing each color code in a bitmap format;
A color code is identified from the run-length data, the number of bits of the length portion corresponding to the identified color code is identified from the base value table, and the run-length data following the color code for the identified number of bits is A differential run length extracting unit for extracting as a differential run length;
An addition unit for calculating a run length by adding a base value corresponding to the difference run length extracted by the difference run length extraction unit;
A run-length decoding apparatus comprising: a decoding unit that decodes the run-length data into the original image data based on the run length calculated by the adding unit and the color information. A run-length encoding method that encodes original image data into run-length data representing each run by a color code part that stores a color code with a predetermined number of bits and a length part in which the number of bits is determined for each color code Because
The run-length encoding device encodes the original image data with a color code and a run length, generates a run-length code, and calculates a run-length histogram of each run for each color code constituting the original image data. A color code assigning step for extracting a part of a run to which a certain color code is assigned based on the calculated histogram and assigning it to an unused color code;
The run-length encoding apparatus obtains, as the number of bits of the length part, the number of bits that minimizes the encoding result of all the runs for each color code after the assignment by the color code assigning step is completed. A run-length encoding method comprising: an encoding step for encoding. The run-length decoding device converts the run-length data representing each run into the original image data using a color code portion that stores a color code with a predetermined number of bits and a length portion in which the number of bits is determined for each color code. A run-length decoding method for decoding
The length part stores a differential run length obtained by subtracting the base value of each run from the run length of each run,
The base value indicates the minimum run length of each color code,
The run length decoding device comprises:
For each color code, a base value table holding unit that holds a base value table in which the base value and the number of bits of the length part are associated;
A color information holding unit for holding color information for representing each color code by the original image data,
The run-length decoding apparatus specifies a color code from the run-length data, specifies the number of bits of the length portion corresponding to the specified color code from the base value table, and the color for the specified number of bits A differential run length extracting step of extracting the run length data following the code as a differential run length;
An addition step of adding a base value corresponding to the difference run length extracted by the difference run length extraction step, and calculating a run length;
A run-length decoding method comprising: a decoding step of decoding the run-length data into the original image data based on the run length calculated in the adding step and the color information.

Images