JP6081862B2 - Image encoding device - Google Patents

Description

  The present invention relates to an image encoding apparatus that compresses and encodes image data.

  2. Description of the Related Art Conventionally, an image encoding device that compresses and encodes image data for the purpose of reducing a burden on communication is known. In recent years, an encoding standard that can be compressed at a higher compression rate has been proposed (see Non-Patent Document 1).

  FIG. 6 shows a schematic configuration of an image encoding device according to the standard. As shown in FIG. 6, the image encoding apparatus 100 includes a color space conversion unit 110, a frequency conversion unit 120, a quantization unit 130, a DC / LP / HP prediction unit 140, and an adaptive scan unit 150. And a VLC encoding unit 160 and a bit stream output unit 170.

  The color space conversion unit 110 processes the input image data for each pixel (pixel), and converts the original image data into YUV data when the original image data is RGB data, and converts the image data into CMYK data. Converts to YUVK data.

  The frequency conversion unit 120 sequentially processes the image data converted by the color space conversion unit 110 in a raster direction with 4 × 4 pixels as one block, and converts the data of each pixel into a coefficient for each frequency band. Process. In this frequency conversion unit 120, for one block (4 × 4 pixels), 1 × 4 pixels in the upper left of 4 × 4 pixels are blank, and the remaining 15 pixels are HP coefficients. The second layer data in which one pixel at the upper left of the pixel is blank and the remaining 15 pixels are LP coefficients and the third layer data composed of DC coefficients of one pixel are generated. The first layer data, the second layer data, and the third layer data are output to the quantization unit 130.

  The quantization unit 130 performs a process of dividing and quantizing the first layer data, the second layer data, and the third layer data that have been coefficientized by the frequency conversion unit 120 by respective set quantization parameters. The processed first tier data, second tier data, and third tier data are output to the DC / LP / HP prediction unit 140.

  The DC / LP / HP prediction unit 140 uses the coefficient values between adjacent blocks for the first layer data, second layer data, and third layer data for each block input from the quantization unit 130, respectively. Are compared to derive a prediction mode, and a process of subtracting between blocks having high correlation is performed in accordance with an arithmetic expression set for the prediction mode. Then, the processed first layer data, second layer data, and third layer data are output to the adaptive scan unit 150 for each block.

  The adaptive scan unit 150 performs processing for replacing the coefficient data of the first layer data and the second layer data by a predetermined method for each block. This replacement process is performed in order to increase the compression rate at the time of encoding by collecting non-zero coefficients on the upper left side of the block (conversely, those having zero coefficient values are collected on the lower right side).

  As shown in FIG. 7, the adaptive scan unit 150 includes a P / S conversion unit 151, a coefficient data replacement processing unit 152, a non-zero data addition unit 153, a comparison / replacement processing unit 154, and a data storage unit 155. As shown in FIG. 9, the comparison / replacement processing unit 154 includes a selector 156, a register 157, a comparison unit 158, and a replacement unit 159. In the adaptive scanning unit 150, processing as shown in FIG. 8 is executed.

  Hereinafter, processing in the adaptive scanning unit 150 will be described. In the following, the processing of the first layer data will be described, but the same applies to the processing of the second layer data. 10 to 16 are explanatory diagrams for explaining the adaptive scan processing.

  The first layer data output from the DC / LP / HP prediction unit 140 is handled in units of blocks corresponding to 4 × 4 pixels as shown in FIGS. The numbers in FIG. 10 represent pixel positions (coordinate positions) in the block, and the HP coefficients shown in FIG. 11 are shown in FIG. 12 in the data table format.

  In the replacement process, the coefficient data Z (i) quantized by the quantization unit is replaced in the block based on the replacement data P (i) stored in the data storage unit. The coefficient data Z ′ (i) for each block after processing is generated. Thereafter, in the frequency update process for updating the non-zero frequency, the non-zero frequency data F (i) stored in the data storage unit is read, and the coefficient data Z ′ (i) in block units after the replacement process is read out. Is updated to the non-zero frequency data F (i) read out by adding 1 to the non-zero frequency at the non-zero pixel position i, and the updated non-zero frequency The frequency data is F ′ (i). Next, in the replacement process for replacing the replacement data P (i) and the frequency update data F (i), the coefficient data Z ′ (i) at each coordinate position i after the replacement process is non-zero. This is performed using non-zero frequency data F ′ (i) after frequency update processing representing the relationship with frequency and replacement data P (i) for replacing coefficient data Z (i) in the block. The replacement data P (i) is preset according to the prediction mode in the DC / LP / HP prediction unit 140, and the replacement data corresponding to the prediction mode derived in the DC / LP / HP prediction unit 140. P (i) is used. In addition, initial values are set in advance for the replacement data P (i) and the non-zero frequency data F (i). An example of initial values of the replacement data P (i) and the non-zero frequency data F (i) is shown in FIG. These replacement data P (i) and non-zero frequency data F (i) are stored in the data storage unit 155.

  From the DC / LP / HP prediction unit 140, information on the prediction mode and the first layer data are input to the P / S conversion unit 151 by one block. The P / S conversion unit 151 first inputs coefficient data Z (i), which is the first layer data (HP coefficient data) for one block, to the coefficient data replacement processing unit 152. The coefficient data replacement processing unit 152 rearranges and holds the coefficient data Z (i) according to the value of the replacement data P (i) read from the data storage unit 155 (which is an initial value in the case of the first block). After the rearrangement of all the coefficient data of one block from Z (1) to Z (15), the rearranged coefficient data Z ′ (1) to Z ′ (15) is VLC encoded as parallel data. Output to the unit 160. The post-replacement coefficient data Z ′ (i) column of FIG. 14 shows the data after the replacement process. In the example shown in FIG. 14, according to the replacement data P (i), the coefficient data Z (2) (value 11) at the coordinate position 2 before the replacement process is incorporated into the coordinate position 5 of the new block, and the replacement is also performed. Coefficient data Z (8) (value is 9A) at the coordinate position 8 before processing is incorporated into the coordinate position 4 of the new block. This process corresponds to step S2 shown in FIG. 8, and the first layer data generated in this way is shown in FIG.

  Next, after the replacement process, the coefficient data replacement processing unit 152 outputs information on the prediction mode to the non-zero data addition unit 153, and then the first hierarchical data (HP coefficient data) Z ′ (1 ) To Z ′ (15) are sequentially input from the coordinate position 1 to the non-zero data adding unit 153.

  In the non-zero data adding unit 153, in synchronization with the input of the HP coefficient data Z ′ (i) from the coefficient data replacement processing unit 152, the non-zero frequency data of the corresponding coordinate position i is stored from the data storage unit 155. When F (i) is read and the HP coefficient data input from the coefficient data replacement processing unit 152 is non-zero, 1 is added to the read non-zero frequency data F (i), and the HP coefficient data is non-zero. If not, the read non-zero frequency data F (i) is output as it is to the comparison / replacement processing unit 154 as non-zero frequency data F ′ (i).

  This process corresponds to steps S6 to S8 shown in FIG. 8, and FIG. 14 shows data after the addition process. In the example shown in FIG. 14, since the coefficient data Z ′ (4) at the coordinate position 4 is 9A and the coefficient data Z ′ (5) at the coordinate position 5 is 11 and non-zero, the non-zero frequency data at the coordinate position 4 1 is added to F (4) = 28 and the updated non-zero frequency data F ′ (4) becomes 29, and 1 is added to the non-zero frequency data F (5) = 24 at the coordinate position 5 and updated. The non-zero frequency data F ′ (5) is 25. The non-zero data addition unit 153 outputs the updated non-zero frequency data F ′ (i) of the coordinate position i = 1 to 15 thus added to the comparison / replacement processing unit 154 for each coordinate position i. .

  The comparison / replacement processing unit 154 compares the non-zero frequency data F ′ (i−1) and F ′ (i) before and after being output from the non-zero data addition unit 153, and F ′ (i) is F ′ ( If it is larger than i−1), F ′ (i−1) and F ′ (i) are replaced, and F ″ (i−1) = F ′ (i), F ″ (i) = F The process of '(i-1) and the replacement data P (i-1) and P (i) before and after read from the data storage unit 155 are replaced, and P' (i-1) = P (i ), P ′ (i) = P (i−1).

  Specifically, in FIG. 9, when F ′ (1) and P (1) at the coordinate position 1 are input to the lower input port of the comparison unit 158, the selector 156 performs the condition of i = 1. The selected F ′ (1) and P (1) are temporarily stored in the register 157, respectively.

  Next, when F ′ (2) and P (2) at the coordinate position 2 are input, the comparison unit 158 reads F ′ (1) and P (1) from the register 157, and F ′ (1) and non-zero. F ′ (2) output from the data adder 153 is compared, and if F ′ (1) <F ′ (2), F ′ (1), F ′ (1), F ′ (2), P (1), and P (2) are output to the replacement unit 159. Then, the replacement unit 159 outputs F ′ (2) as F ″ (1) at the coordinate position 1 and P (2) as P ′ (1) at the coordinate position 1 to the data storage unit 155 to output the coordinate position 2 F ″ (2) as F ″ (2) and P (1) as P ′ (2) are temporarily stored in the register 157 via the selector 156, respectively.

  On the other hand, if F ′ (1) ≧ F ′ (2), the comparison unit 158 performs F ′ (1), F ′ (2), P (1) together with the replacement non-execution signal (R (2) = 0). And P (2) are output to the replacement unit 159, and the replacement unit 159 stores F ′ (1) and P (1) at the coordinate position 1 as F ″ (1) and P ′ (1) in the data storage unit 155. And F ′ (2) and P (2) at the coordinate position 2 are temporarily stored in the register 157 via the selector 156. Thereafter, F ′ (3-15) at the remaining coordinate position i = 3 to 15 And P (3 to 15), the same processing is repeated.

  The above processing corresponds to steps S10 to S12 shown in FIG. 8, and FIG. 15 shows data after the replacement processing. The example shown in FIG. 15 is obtained by replacing the one shown in FIG. 14, and F ′ (4)> F ′ (3) in the state before the replacement shown in FIG. P (3) at position 3 and P (4) at coordinate position 4 are replaced, and F ′ (3) at coordinate position 3 and F ′ (4) at coordinate position 4 are replaced.

  The F (i) and P (i) stored in the data storage unit 155 are updated by F ″ (i) and P ′ (i) output from the comparison / replacement processing unit 154, respectively.

  In the adaptive scanning unit 150, the above-described processing is performed, and the first hierarchical data in which the coefficient data is replaced is generated in accordance with the non-zero frequency and preset replacement data. Similar processing is performed on the second layer data, and second layer data in which the coefficient data is replaced is generated. The third layer data is output to the VLC encoder 160 as it is.

  The VLC encoder 160 performs entropy encoding on the first layer data, the second layer data, and the third layer data processed by the adaptive scan unit 150, generates a code stream, and generates the bit The bit stream output unit 170 outputs a stream to the stream output unit 170. The bit stream output unit 170 collects the input bit streams for one image data as one, forms one file, and outputs the formed file to the outside.

JPEG XR Recommendation ITU-T T.832: Information technology -JPEG XR image coding system-Image coding specification. [Online] .Recommendation T.832 (01/12). Approval 2012-01-13. Status: In force. Retrieved from the Internet: <URL: http://www.itu.int/rec/T-REC-T.832-201201-I/en>.

  By the way, the amount of data of image data to be compressed has increased rapidly due to the improvement in resolution accompanying technological advancement. In the conventional image encoding apparatus 100, the processing takes a long time due to the increase in data amount. Therefore, further efficiency is required.

  In particular, the adaptive scan unit 150 compares the preceding and following non-zero frequencies F ′ (i−1) and F ′ (i), and then compares the non-zero frequencies F ′ (i) and the replacement data P (i). Since the replacement process is a process in which the process is repeated as many times as the number of coefficient data, the process must be repeated as many times as the total number of first layer data and second layer data, The process took a long time. For this reason, the processing of the conventional adaptive scanning unit 150 has become one of the bottleneck factors that reduce the processing efficiency of the image coding apparatus 100. The one-cycle processing is performed within one clock. In the electronic circuit using the register 157, the clock is used as the data read timing. Therefore, the processing in the conventional adaptive scan unit 150 is performed. The time required a time multiple of one clock cycle (= 1 clock × cycle number).

  The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide an image encoding device capable of improving the overall processing efficiency by increasing the processing efficiency of the adaptive scanning unit. And

In order to solve the above-described problem, an image encoding device according to the present invention includes:
A frequency conversion unit that converts the frequency of image data into a block unit of a predetermined number of pixels to generate a coefficient;
A quantization unit that quantizes the n coefficient data in units of blocks that have been coefficientized by the frequency conversion unit;
Non-zero frequency data F (i) representing the relationship between the pixel position i of the block and the frequency at which the coefficient data Z (i) at each pixel position i was non-zero, and coefficient data Z (i ) Is replacement data P (i) for replacing the current pixel position i and the replacement data representing the relationship between the pixel position i ′ to which the coefficient data Z (i) of the pixel position i should be incorporated by replacement. An adaptive scanning unit that uses P (i) to replace the block unit coefficient data Z (i) quantized by the quantization unit in the block;
An image encoding device including at least an encoding unit that encodes and compresses the coefficient data Z ′ (i) in block units after the replacement process;
The adaptive scan unit includes:
A data storage unit for storing the non-zero frequency data F (i) and replacement data P (i);
A coefficient data replacement processing unit that performs a replacement process on the coefficient data Z (i) quantized by the quantization unit in the block based on the replacement data P (i) stored in the data storage unit;
The non-zero frequency data F (i) stored in the data storage unit is read out, and it is confirmed whether or not the block-unit coefficient data Z ′ (i) after the replacement processing is non-zero. A frequency update processing unit that adds 1 to the non-zero frequency at the pixel position i that is zero, updates the read non-zero frequency data F (i), and sets the updated non-zero frequency data F ′ (i) as an update, ,
Using the non-zero frequency data F ′ (i) updated by the frequency update processing unit, F ′ (i−1) and F ′ (i) are sequentially compared, and F ′ (i−1) and A replacement information generation processing unit for generating replacement information R (i) indicating whether or not F ′ (i), P (i−1) and P (i) should be replaced;
The replacement data P (i) stored in the data storage unit is read, and F ′ (updated by the frequency update processing unit based on the replacement information R (i) generated by the replacement information generation processing unit. i-1) and F ′ (i) and the read P (i-1) and P (i) are replaced, and the non-zero frequency data F ″ (i) and replacement data P after the replacement are processed. '(I) relates to an image encoding device including a replacement processing unit for updating non-zero frequency data F (i) and replacement data P (i) stored in the data storage unit, where n is It is an integer greater than or equal to 2, and said i is an integer of 1-n.

In the image encoding device of the present invention,
The replacement processing unit includes (n-1) replacers,
Each replacer has three input ports A, B, C and two output ports D, E,
The output port E of the jth replacer is sequentially connected to the input port A of the (j + 1) th replacer,
The first to the input port A of the replacement unit, the non-zero frequency data F '(1) and replacement data P (1) is input to the input port B of the j th of said replacement unit, the non-zero frequency data F '(j + 1) and replacement data P (j + 1) are input, j-th wherein the input port C of the replacement unit of the with replacement information R (j + 1) are input, each substituted instrument, the input The replacement process is executed according to the replacement information R (j + 1) input from the port C, and the non-zero frequency data F ″ (j) and replacement data P ′ (j) after the replacement process are output ports D of the respective replacers. The non-zero frequency data F ″ (n) and replacement data P ′ (n) after the replacement processing are output from the output port E of the (n−1) th replacer. However, said n is an integer greater than or equal to 2, and said i is an integer of 1-n. J is an integer of 1 to (n-1).

  According to the image coding apparatus according to the present invention having the above-described configuration, first, image data for one image is input to the frequency conversion unit. Then, in the frequency conversion unit, the image data is frequency converted and coefficientized in units of a block of a predetermined number of pixels, and then the coefficient data of the block unit thus quantized is quantized in the quantization unit, and is converted into DC / LP / The HP prediction unit 140 derives a prediction mode and performs a process of subtracting between highly correlated blocks, and then inputs the prediction mode to the adaptive scanning unit.

  In the adaptive scanning unit, the non-zero frequency data F (i) representing the relationship between the pixel position i of the block and the frequency at which the coefficient data Z (i) at each pixel position i is non-zero, and within the block Replacement data P (i) for replacing the coefficient data Z (i) between the current pixel position i and the pixel position i ′ where the coefficient data Z (i) at the pixel position i is to be incorporated by replacement. Using the replacement data P (i) representing the relationship, the quantized block unit coefficient data Z (i) is replaced in the block, and the block unit coefficient data Z ′ (i) after replacement is performed. ) Is encoded and compressed in the encoding unit.

And the processing in the adaptive scanning unit is as follows:
It is checked whether or not the quantized block-unit coefficient data Z (i) is non-zero, and 1 is added to the non-zero frequency of the non-zero pixel position i to obtain non-zero frequency data F (I) is updated and the updated non-zero frequency data F ′ (i) is processed by the frequency update processing unit;
Using the non-zero frequency data F ′ (i) updated by the frequency update processing unit, F ′ (i−1) and F ′ (i) are sequentially compared, and F ′ (i−1) and F ′ (i), processing by the replacement information generation processing unit that generates replacement information R (i) indicating whether or not P (i−1) and P (i) should be replaced;
Based on the replacement information R (i) generated by the replacement information generation process, the replacement process of F ′ (i−1) and F ′ (i), P (i−1) and P (i) is performed. Processing by the replacement processing unit;
This processing includes processing by the coefficient data replacement processing unit that replaces the coefficient data Z (i) in the block based on the replacement data P (i).

  As before, after comparing the preceding and following non-zero frequency data F ′ (i−1) and F ′ (i), F ′ (i−1), F ′ (i), and P (i−1) And P (i) are replaced by the number of coefficient data Z (i), the processing cycle must be repeated. However, as in the present invention, the comparison process and the replacement process are separated, and the comparison process is performed. After the replacement information is generated in advance, by performing replacement processing based on the replacement information, the processing in the frequency update processing unit, replacement information generation processing unit, and replacement processing unit is performed by a plurality of coefficient data Z (i) groups. Can be processed in one cycle.

  Note that this one-cycle processing is performed within one clock of a clock used as processing timing.

  According to the image coding apparatus according to the present invention having the above-described configuration, the replacement information is generated by the replacement information using the replacement information generated in advance by the replacement information generation unit. Processing efficiency in one cycle can be improved while suppressing deterioration of the operating frequency characteristics of the replacement processing unit.

In the image encoding device of the present invention,
Each of the frequency update processing unit, the replacement information generation processing unit, and the replacement processing unit has a plurality of cycles in which n coefficient data Z (i) groups constituting the one block are set to m in one cycle. It is configured to process repeatedly. However, n is an integer of 3 or more, and m is an integer of 2 or more, and is an integer smaller than the n. The i is an integer from 1 to n.

In the image encoding device of the present invention,
The n coefficient data Z (i) group constituting one block is configured to repeatedly process a plurality of cycles, with the number of processes in one cycle being m.
The replacement processing unit includes (m−1) number of replacers from the first to (m−1) th, where m is the number of coefficient data Z (i) groups processed in one cycle,
Each replacer has three input ports A, B, C and two output ports D, E,
The output port E of the kth replacer is sequentially connected to the input port A of the k + 1th replacer,
The input port A of the first shifter, a non-zero frequency data F '(a) and replacement data P (a) is input to the input port B of the k-th of said replacement unit, the non-zero frequency data F '(a + 1) and replacement data P (a + 1) is input, k-th wherein the input port C of the replacement unit of the with replacement information R (a + 1) is input, at each substitutable instrument, the input The replacement process is executed according to the replacement information R (a + 1) input from the port C, and the non-zero frequency data F ″ (a) and replacement data P ′ (a) after the replacement process are output to the output port D of each replacer. And the non-zero frequency data F ″ (a + 1) and replacement data P ′ (a + 1) after the replacement process are output from the output port E of the (m−1) th replacer. It is characterized by that. However, said n is an integer greater than or equal to 3, and said i is an integer of 1-n. The m is an integer of 2 or more and is an integer smaller than the n, and the k is an integer of 1 to (m−1). Further, a is a = [k + (r−1) × (m−1)], and r is the number of repetitions.

  According to the image coding apparatus according to the present invention having the above-described configuration, the processing of coefficient data of one block can be divided into a plurality of cycles by making the number of replacement processes in one cycle variable.

  According to the image coding apparatus according to the present invention having the above-described configuration, the replacement information is generated by the replacement information using the replacement information generated in advance by the replacement information generation unit. It is possible to minimize the number of replacement processors that can be processed in each cycle divided into cycles.

  Therefore, according to the adaptive scan unit of the present invention, the processing time is shorter than that of the conventional adaptive scan unit, and the replacement process can be performed efficiently.

  As described above, according to the present invention, the processing time of the adaptive scanning unit can be shortened compared to the conventional case, and the replacement process can be performed efficiently. As a result, the overall processing speed of the image encoding device can be increased compared to the conventional case, and an efficient encoding process can be realized.

It is the block diagram which showed schematic structure of the image coding apparatus which concerns on one Embodiment of this invention. It is the block diagram which showed the structure of the adaptive type scan part which concerns on this embodiment. It is explanatory drawing for demonstrating the process in the replacement information generation part which concerns on this embodiment. It is explanatory drawing for demonstrating the process in the replacement information generation part which concerns on this embodiment. It is the block diagram which showed the structure of the replacement process part which concerns on this embodiment. It is the block diagram which showed schematic structure of the conventional image coding apparatus. It is the block diagram which showed the structure of the conventional adaptive type scanning part. It is the flowchart which showed the process sequence in the conventional adaptive type scanning part. It is the block diagram which showed the structure of the conventional comparison and replacement process part. It is explanatory drawing for demonstrating the pixel position within the block of the conventional adaptive scanning process. It is explanatory drawing for demonstrating an example of the HP coefficient within the block of the conventional adaptive type scanning process. It is explanatory drawing for demonstrating an example which made the HP coefficient of the conventional adaptive type scanning process into the data table format. It is explanatory drawing for demonstrating an example of the initial value of the replacement data and the non-zero frequency data of the conventional adaptive scanning process. It is explanatory drawing for demonstrating replacement | exchange of the coefficient data of a conventional adaptive type scanning process, and update of a non-zero frequency. It is explanatory drawing for demonstrating the replacement data and the non-zero frequency data after the replacement of the conventional adaptive scanning process. It is explanatory drawing for demonstrating an example of the HP coefficient after substitution within the block of the conventional adaptive scan process.

  Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the image encoding device 1 according to this embodiment.

  As shown in FIG. 1, the image encoding device 1 of this example includes an adaptive scanning unit 10 having a configuration different from that of the adaptive scanning unit 150 in the conventional image encoding device 100 of FIG. 6. Only in this respect, the configuration of the conventional image encoding device 100 is different. Therefore, the other components are denoted by the same reference numerals as those in the image encoding device 100, and detailed description thereof is omitted.

  As shown in FIG. 2, the adaptive scanning unit 10 includes a coefficient data replacement processing unit 20, a frequency update processing unit 30, a replacement information generation processing unit 40, a replacement processing unit 50, and a data storage unit 60.

  In the adaptive scan unit 10 as well, the first layer data and the second layer data output from the DC / LP / HP prediction unit 140 are shown in FIG. 10 and FIG. Processing is performed in units of blocks corresponding to 4 × 4 pixels as shown, and processing for replacing the coefficient data Z (i) in the block is performed. In the following, the processing of the first layer data will be described, but the same applies to the processing of the second layer data.

  The replacement process replaces the non-zero frequency data F (i) representing the relationship with the frequency where the coefficient data Z (i) at each coordinate position i is non-zero, and the coefficient data Z (i) in the block. The replacement data P (i) is used. The replacement data P (i) is set in advance according to the prediction mode in the DC / LP / HP prediction unit 140, and the replacement data P () corresponding to the prediction mode derived in the DC / LP / HP prediction unit 140. i) is used. Further, the replacement data P (i) and the non-zero frequency data F (i) are set in advance with initial values. These replacement data P (i) and non-zero frequency data F (i) are stored in the data storage unit 60.

  The DC / LP / HP prediction unit 140 inputs the information related to the derived prediction mode and the first layer data Z (i) to the coefficient data replacement processing unit 20 for each block, and the coefficient data replacement processing unit 20 The coefficient data Z (i) is rearranged according to the value of the replacement data P (i) read from the data storage unit 60 (which is the initial value in the case of the first block), and is output to the VLC encoding unit 160.

  Thereafter, the coefficient data replacement processing unit 20 first outputs information on the prediction mode to the frequency update processing unit 30, and then the coordinates after replacement processing in the first layer data (HP coefficient data) for one block. Coefficient data Z ′ (i = 1 to 8) at positions 1 to 8 is input to the frequency update processing unit 30, and then the coefficient data Z ′ (i = 8 to 15) at coordinate positions 8 to 15 after the replacement process is input. Input to the frequency update processing unit 30. Thereafter, similarly, the coefficient data replacement processing unit 20 inputs the data in block units input from the DC / LP / HP prediction unit 140 to the frequency update processing unit 30.

  The frequency update processing unit 30 is synchronized with the input of coefficient data Z ′ (i = 1 to 8) after the replacement processing of the coordinate positions 1 to 8 from the coefficient data replacement processing unit 20. 60, the non-zero frequency data F (i = 1 to 8) of the coordinate positions 1 to 8 is read, and the non-zero frequency corresponding to the coordinate position i where the coefficient data Z ′ (i = 1 to 8) is non-zero. A process of adding 1 to the data F (i) to generate updated non-zero frequency data F ′ (i) and outputting the generated non-zero frequency data F ′ (i) to the replacement information generation processing unit 40 I do.

  This process is performed by using a circuit in which adders are arranged in parallel. In one cycle process, eight non-zero frequency data F (i) are updated, and the frequency update processing unit 30 is input next. Similarly, the non-zero frequency data F (i = 8 to 15) is updated based on the coefficient data 'Z (i = 8 to 15) after the replacement processing of the coordinate positions 8 to 15, and the non-zero frequency data F'. (I) is generated, and the generated non-zero frequency data F ′ (i) is output to the replacement information generation processing unit 40. Then, the frequency update processing unit 30 repeats this processing for each block and performs processing for one image data. The updated non-zero frequency data F ′ (i = 1 to 15) is the same as that illustrated in FIG. 14.

  The replacement information generation processing unit 40 generates replacement information R (i) based on the eight updated non-zero frequency data F ′ (i = 1 to 8) output from the frequency update processing unit 30. This replacement information R (i) compares the preceding and following non-zero frequencies F ′ (i−1) and F ′ (i), and when F ′ (i) is larger than F ′ (i−1). In a replacement processing unit 50 described later, F ′ (i−1) and F ′ (i) are replaced, and F ″ (i−1) = F ′ (i), F ″ (i) = F ′. (I-1) and P (i-1) and P (i) are replaced, and P '(i-1) = P (i), P' (i) = P (i- This is information for performing the processing 1).

Specifically, the replacement information includes a flag, and the flag is generated under the following conditions as shown in FIG.
1) For F ′ (i)> F ′ (i−1), if this condition is true, “Y” (meaning “Yes”, and so on).
2) For F ′ (i) = F ′ (i−1), “Y” if this condition is true.
3) If the condition of F ′ (i)> F ′ (i−1) is true, the replacement information R (i) is set to 1.
4) Even when the condition of F ′ (i)> F ′ (i−1) is false in the determination of the coordinate position, the condition of F ′ (i) = F ′ (i−1) is true, and If the replacement information R (i) at the previous coordinate position is 1, the replacement information R (i) is set to 1.
5) In cases other than the above 3) and 4), the replacement information R (i) is set to 0.

  As described above, when F ′ (i) is larger than F ′ (i−1), the substitution between F ′ (i−1) and F ′ (i), and P (i−1) and P As shown in FIG. 3, when F ′ (2)> F ′ (1) and F ′ (2) to F ′ (5) are equivalent, as shown in FIG. 3, F ′ (1 ) To F ′ (5) need to be sequentially replaced, so that the replacement information R (i) at the coordinate positions 2 to 5 is set to 1 under the conditions described in 3) and 4) above.

  The replacement information generation processing unit 40 generates replacement information R (i) as described above. The replacement information generation processing unit 40 generates replacement information R (i = 2 to 8) and non-generation information generated from the eight non-zero frequency data F ′ (i = 1 to 8) output from the frequency update processing unit 30. Zero frequency data F ′ (i = 1 to 8) is output to the replacement processing unit 50. The replacement data P (i = 1 to 8) read from the data storage unit 60 is also output to the replacement processing unit 50, and the replacement processing unit 50 performs replacement based on the replacement information R (i = 2 to 8). The non-zero frequency data F ″ (i = 1 to 8) and the replacement data P ′ (i = 1 to 8) are output to the data storage unit 60.

  Then, the replacement information generation processing unit 40 performs the same processing on the eight non-zero frequency data F ′ (i = 8 to 15) output from the frequency update processing unit 30 to perform replacement information R (i = 9). To 15) (see FIG. 4), replacement information R (i = 9 to 15) and non-zero frequency data F ′ (i = 9 to 15), replacement data P (i = 9 to 15) are output to the replacement processing unit 50. The replacement processing unit 50 replaces the non-zero frequency data F ″ (i = 9 to 15) and the replacement based on the replacement information R (i = 9 to 15). Data P ′ (i = 9 to 15) is output to the data storage unit 60.

  In this way, the replacement information generation processing unit 40 generates replacement information R (i = 2 to 15) for one block in two cycles, and repeats this processing in units of blocks to process one image data. Do.

  As shown in FIG. 5, the replacement processing unit 50 includes seven replacement units arranged in parallel, that is, a first replacement unit 61, a second replacement unit 62, a third replacement unit 63, a fourth replacement unit 64, A fifth replacer 65, a sixth replacer 66, a seventh replacer 67, a register 68, and a selector 69 are provided. The first to seventh replacers 61 to 67 are respectively input port A to which non-zero frequency data F ′ (i) and replacement data P (i) are input, and non-zero frequency data F ′ (i + 1) and replacement. Input port B to which data P (i + 1) is input, input port C to which replacement information R (i + 1) is input, non-zero frequency data F ″ (i) and replacement data P ′ (i) after replacement An output port D that outputs data and an output port E that outputs non-zero frequency data F ″ (i + 1) and replacement data P ′ (i + 1) after replacement are provided.

  The output port E that outputs the non-zero frequency data F ″ (i + 1) and the replacement data P ′ (i + 1) of the first replacer 61 is connected to the non-zero frequency data F ′ (i) of the second replacer 62 and the replacement. Similarly, the output connected to the input port A for inputting the data P (i) and outputting the non-zero frequency data F ″ (i + 1) and the replacement data P ′ (i + 1) of the second to sixth replacers 62 to 66. The port E is connected to an input port A for inputting the non-zero frequency data F ′ (i) and replacement data P (i) of the third to seventh replacers 63 to 67, respectively. Thus, the non-zero frequency data F ″ (i + 1) and replacement data P ′ (i + 1) output from the first to sixth replacers 61 to 66 are respectively sent to the second to seventh replacers 62 to 67. Are input as non-zero frequency data F ′ (i) and replacement data P (i).

  Further, the output port D that outputs the non-zero frequency data F ″ (i) and the replacement data P ′ (i) of each of the first to seventh replacers 61 to 67 is connected to the table data storage unit 60, and the seventh replacement An output port E that outputs the non-zero frequency data F ″ (i + 1) and replacement data P ′ (i + 1) of the device 67 is also connected to the table data storage unit 60. The output port E that outputs the non-zero frequency data F ″ (i + 1) and replacement data P ′ (i + 1) of the seventh replacer 67 is also connected to the register 68, and the non-zero frequency data F ″ (8 ) And replacement data P ′ (8) only are output to the register 68. The output port of the register 68 is connected to the input port A for inputting the non-zero frequency data F ′ (i) and the replacement data P (i) of the first replacer 61 via the selector 69, and the first time of each block The non-zero frequency data F ′ (1) output from the replacement information generation processing unit 40 and the replacement data P (1) read from the data storage unit 60 are sent to the first replacement unit 61 under the selection condition of i = 1. The non-zero frequency data F ″ (8) and replacement data P ′ (8) held in the register 68 are input to the input port A of the first replacer 61 under other selection conditions.

  Thus, in the replacement processing unit 50, first, the replacement information R (i = 2 to 8) and the non-zero frequency data F ′ (i = 1 to 8), the data storage unit 60 from the replacement information generation processing unit 40. The replacement data P (i = 1 to 8) is input to the first to seventh replacers 61 to 67, respectively. That is, F ′ (1), P (1), F ′ (2), P (2) and R (2) are input to the first replacer 61, and F ′ (3), P (3) and R (3) is input to the second replacer 62, and similarly, F ′ (i = 4 to 8), P (i = 4 to 8) and R are respectively input to the third to seventh replacers 63 to 67. (I = 4 to 8) is input.

  Then, in the first replacer 61, based on the replacement information R (2), replacement processing of the non-zero frequency data F ′ (1) and F ′ (2) and replacement data P (1) and P (2) The replacement process is performed. That is, if the replacement information R (2) is 1, F ′ (2) is output as F ″ (1), P (2) is output as P ′ (1), and the second replacer 62 F ′ (1) is input as F ′ (2) and P (1) is input as P (2). On the other hand, if the replacement information R (2) is 0, F ′ (1) F '(1) is output, P (1) is output as P' (1), and F '(2) is input as F' (2) to the second replacer 62, and P (2) P (2) is input as

  In the second to seventh replacers 62 to 67, the same processing is performed. From the second to seventh replacers 62 to 67, F ″ (i = 2 to 8) and P ′ (i = 2 to 8), respectively. Then, F ″ (i = 1 to 8) and P ′ (i = 1 to 8) output in this way are input to the data storage unit 60. Further, F ″ (8) and P ′ (8) are also input to the register 68 and held in this register 68 for the next processing.

  Next, replacement information R (i = 9 to 15), non-zero frequency data F ′ (i = 9 to 15), and replacement data P (i = 9 to 15) are input to the first to seventh replacers 61 to 67, respectively. Further, F ″ (8) and P ′ (8) are input from the register 68 to the first replacer 61 as F ′ (8) and P (8) through the selector 69.

  Then, in the first to seventh replacers 61 to 67, the same replacement process as described above is performed, and F ″ (i = 8 to 15) and P ′ (i = 8 to 15) after replacement are output. Is input to the data storage unit 60. In the data storage unit 60, F (i) and P are input by F ″ (i = 1 to 15) and P ′ (i = 1 to 15) input from the replacement processing unit 50. (I) is updated.

  In this way, the replacement processing unit 50 performs replacement processing for one block in two cycles, and repeats this processing in units of blocks to perform processing on one image data. F ″ (i) and P ′ (i) after one block processing are as illustrated in FIG.

  According to the image coding apparatus 1 of the present example having the above configuration, first, image data for one image is input from the outside to the color space conversion unit 110, and the image data is subjected to color conversion processing. Are input to the frequency converter 120.

  The frequency conversion unit 120 sequentially processes 4 × 4 pixels as one block in the raster direction, converts the data of each pixel into a coefficient for each frequency band, and processes the first layer data and the second layer after processing. The data and the third layer data are output to the quantization unit 130 for each block.

  The quantization unit 130 performs a process of dividing the first layer data, the second layer data, and the third layer data that have been coefficientized by the frequency conversion unit 120 by dividing by the set quantization parameter, respectively, The subsequent first layer data, second layer data, and third layer data are output to the DC / LP / HP prediction unit 140 for each block, and the DC / LP / HP prediction unit 140 similarly uses the first layer for each block. The prediction mode is derived for the data, the second tier data, and the third tier data, and the process of subtracting between the highly correlated blocks is performed, and the first tier data, the second tier data, and the third tier data after processing are blocked. Each time, the data is output to the adaptive scan unit 10.

  The adaptive scanning unit 10 processes the first layer data and the second layer data input from the DC / LP / HP prediction unit 140 in units of blocks corresponding to 4 × 4 pixels, and generates coefficient data Z in the block. The process of replacing (i) is performed, and the processed first layer data and second layer data are output to the VLC encoding processing unit 160. After replacement processing, replacement data P ′ (i = 1 to 8) for performing replacement processing is generated, and replacement data P (i = 1 to 8) is updated for processing of the next block, and P ′ ( i = 8 to 15) is generated by executing two cycles of processing for updating the replacement data P (i = 8 to 15) for processing of the next block. Therefore, according to the adaptive scanning unit 10 of this example, in order to generate the replacement data P ′ (i), it is necessary to repeat the same number of cycles as the number of coefficient data, compared with the conventional adaptive scanning unit 150. Thus, the processing time is short and the replacement process can be performed efficiently. In this embodiment, the replacement data P (i = 1 to 15) is updated for the processing of the next block after the coefficient replacement processing. Although it is different from the specification of Non-Patent Document 1, a configuration may be adopted in which the replacement processing of the coefficients of the same block is performed after the replacement data P (i = 1 to 15) is updated. Note that the third layer data input from the DC / LP / HP prediction unit 140 is output to the VLC encoding processing unit 160 as it is.

  The VLC encoding processing unit 160 performs entropy encoding processing on the first layer data, the second layer data, and the third layer data output from the adaptive scan unit 10, generates a code stream, and generates a bit stream. The output is performed by the output unit 170. The bit stream output unit 170 forms a single file by combining the input code streams for one image data, and outputs the formed file to the outside.

  As described in detail above, according to the image encoding device 1 of this example, the replacement data P ′ (i) for performing the process of replacing the coefficient data Z (i) in the adaptive scan unit 10 is P Since it is generated by executing two cycles of processing for generating '(i = 1 to 8) and processing for generating P' (i = 8 to 15), compared to the conventional adaptive scanning unit 150 Therefore, the processing time is short and the replacement process can be performed efficiently. As a result, the processing speed of the entire image encoding device 1 can be increased compared to the conventional case, and an efficient encoding process is realized. can do.

  Note that one cycle of processing is performed within one clock of the clock used as processing timing, and therefore, in this example, one block of adaptive scan is executed in two clocks.

  As mentioned above, although one Embodiment of this invention was described, the specific aspect which this invention can take is not limited to this at all. For example, in the above example, the generation process of the replacement data P ′ (i) in the adaptive scanning unit 10 is generated in two cycles. However, the present invention is not limited to this, and the replacement data P in three or more cycles. '(I) may be generated, and if possible, replacement data P' (i) may be generated by processing 15 blocks of data in one cycle.

  In this case, although it is necessary to set the number of coefficient data Z (i) groups to be processed according to the number of processing cycles, the number of replacers in the replacement processing unit 50 is the coefficient data Z processed in one cycle. (I) It is set according to the number of groups. That is, the number of replacers is (the number of coefficient data Z (i) groups processed in one cycle minus 1). However, for the purpose of the present invention, it is important to process the number of cycles smaller than the number of coefficient data.

DESCRIPTION OF SYMBOLS 1 Image coding apparatus 10 Adaptive scanning part 20 Coefficient data replacement | exchange part 30 Frequency update processing part 40 Replacement information generation processing part 50 Replacement | exchange part 60 Data memory | storage part 61 1st substitution device 62 2nd substitution device 63 3rd substitution device 64 Fourth replacement unit 65 Fifth replacement unit 66 Sixth replacement unit 67 Seventh replacement unit 68 Register 69 Selector 110 Color space conversion unit 120 Frequency conversion unit 130 Quantization unit 140 DC / LP / HP prediction unit 150 Adaptive scan unit 151 P / S conversion section 152 Coefficient data replacement processing section 153 Non-zero data addition section 154 Comparison / replacement processing section 155 Data storage section 156 Selector 157 Register 158 Comparison section 159 Replacement section 160 VLC encoding section 170 Bit stream output section

Claims (4)

A frequency conversion unit that converts the frequency of image data into a block unit of a predetermined number of pixels to generate a coefficient;
A quantization unit that quantizes the n coefficient data in units of blocks that have been coefficientized by the frequency conversion unit;
Non-zero frequency data F (i) representing the relationship between the pixel position i of the block and the frequency at which the coefficient data Z (i) at each pixel position i was non-zero, and coefficient data Z (i ) Is replacement data P (i) for replacing the current pixel position i and the replacement data representing the relationship between the pixel position i ′ to which the coefficient data Z (i) of the pixel position i should be incorporated by replacement. An adaptive scanning unit that uses P (i) to replace the block unit coefficient data Z (i) quantized by the quantization unit in the block;
An image encoding device including at least an encoding unit that encodes and compresses the coefficient data Z ′ (i) in block units after the replacement process;
The adaptive scan unit includes:
A data storage unit for storing the non-zero frequency data F (i) and replacement data P (i);
A coefficient data replacement processing unit that performs a replacement process on the coefficient data Z (i) quantized by the quantization unit in the block based on the replacement data P (i) stored in the data storage unit;
The non-zero frequency data F (i) stored in the data storage unit is read out, and it is confirmed whether or not the coefficient data Z ′ (i) of the block unit after the replacement process is non-zero, and the non-zero A frequency update processing unit that adds 1 to the non-zero frequency at the pixel position i and updates the read non-zero frequency data F (i) to be updated non-zero frequency data F ′ (i);
Using the non-zero frequency data F ′ (i) updated by the frequency update processing unit, F ′ (i−1) and F ′ (i) are sequentially compared, and F ′ (i−1) and A replacement information generation processing unit for generating replacement information R (i) indicating whether or not F ′ (i), P (i−1) and P (i) should be replaced;
The replacement data P (i) stored in the data storage unit is read, and F ′ (updated by the frequency update processing unit based on the replacement information R (i) generated by the replacement information generation processing unit. i-1) and F ′ (i) and the read P (i-1) and P (i) are replaced, and the non-zero frequency data F ″ (i) and replacement data P after the replacement are processed. An image encoding device comprising: a non-zero frequency data F (i) and replacement data P (i) stored in the data storage unit in (i) and a replacement processing unit P (i) However, n is an integer of 2 or more, and i is an integer of 1 to n.   Each of the frequency update processing unit, the replacement information generation processing unit, and the replacement processing unit has a plurality of cycles in which n coefficient data Z (i) groups constituting the one block are set to m in one cycle. The image coding apparatus according to claim 1, wherein the image coding apparatus is configured to repeatedly perform processing. However, said n is an integer greater than or equal to 3, and said m is an integer greater than or equal to 2, and is an integer smaller than said n. The i is an integer from 1 to n. The replacement processing unit includes (n-1) replacers,
Each replacer has three input ports A, B, C and two output ports D, E,
The output port E of the jth replacer is sequentially connected to the input port A of the (j + 1) th replacer,
The first to the input port A of the replacement unit, the non-zero frequency data F '(1) and replacement data P (1) is input to the input port B of the j th of said replacement unit, the non-zero frequency data F '(j + 1) and replacement data P (j + 1) are input, j-th wherein the input port C of the replacement unit of the with replacement information R (j + 1) are input, each substituted instrument, the input The replacement process is executed according to the replacement information R (j + 1) input from the port C, and the non-zero frequency data F ″ (j) and replacement data P ′ (j) after the replacement process are output ports D of the respective replacers. And the non-zero frequency data F ″ (n) and replacement data P ′ (n) after the replacement process are output from the output port E of the (n−1) th replacer. The image coding apparatus according to claim 1. However, said n is an integer greater than or equal to 2, and said i is an integer of 1-n. The j is an integer from 1 to (n-1). The n coefficient data Z (i) group constituting one block is configured to repeatedly process a plurality of cycles, with the number of processes in one cycle being m.
The replacement processing unit includes (m−1) number of replacers from the first to (m−1) th, where m is the number of coefficient data Z (i) groups processed in one cycle,
Each replacer has three input ports A, B, C and two output ports D, E,
The output port E of the kth replacer is sequentially connected to the input port A of the k + 1th replacer,
The input port A of the first shifter, a non-zero frequency data F '(a) and replacement data P (a) is input to the input port B of the k-th of said replacement unit, the non-zero frequency data F '(a + 1) and replacement data P (a + 1) is input, k-th wherein the input port C of the replacement unit of the with replacement information R (a + 1) is input, at each substitutable instrument, the input The replacement process is executed according to the replacement information R (a + 1) input from the port C, and the non-zero frequency data F ″ (a) and replacement data P ′ (a) after the replacement process are output to the output port D of each replacer. And the non-zero frequency data F ″ (a + 1) and replacement data P ′ (a + 1) after the replacement process are output from the output port E of the (m−1) th replacer. The image code according to claim 2, Device. However, said n is an integer greater than or equal to 3, and said i is an integer of 1-n. The m is an integer of 2 or more and is an integer smaller than the n, and the k is an integer of 1 to (m−1). The a is a = [k + (r−1) × (m−1)], and the r is the number of repetitions.

Images