JP3715664B2 - Image processing method - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to an image processing method for processing a moving image.
[0002]
[Prior art]
In recent years, international standardization of image data encoding techniques has been actively performed. The JPEG (color still image coding) system has become an international standard as a coding system for color multilevel images. In this method, an input color multi-valued image is divided into blocks, subjected to DCT transform, quantized coefficients, assigned a Huffman code as a quantization result, and encoded. Details are described in "Outline of Color Still Image Coding International Standard System (JPEG) (Part 1) (Technical Explanation)" [Journal of Image Electronics Society, Vol. 20, No. 1].
[0003]
International standardization of moving images is also underway. H.TV for video conference applications The 261 (video coding for communication) method and the MPEG (moving image coding for storage) method using storage media have become standard. These encoding methods are methods in which a basic frame is encoded in the frame and other frames are encoded by motion compensation. The basics of intraframe coding are block division and DCT conversion as in the JPEG method. Also in motion compensation, a motion vector is calculated for each block, a difference from the target block is obtained, a difference block is DCT transformed, a coefficient is quantized, a Huffman code is assigned to a quantization result, and encoding is performed. is there. Details are published in "International Standard for Multimedia Coding" [Maruzen Co., Ltd.].
[0004]
While these standardizations are progressing, LSIs for these technologies are rapidly being realized. JPEG LSIs were first announced in the order of standardization. Later, LSIs that perform JPEG encoding were supplied in large quantities at low cost. As a result, before the MPEG method, which is a moving image encoding method, has been standardized, an application for encoding each frame of a moving image using the JPEG method and storing it on a magnetic disk or the like has become widespread. Hereinafter, data obtained by encoding a moving image by the JPEG method is referred to as Motion JPEG data.
[0005]
These applications are characterized in that editing in frame units is easier and handling is easier than in the MPEG system. On the other hand, the MPEG system is characterized by high coding efficiency because it employs interframe coding.
[0006]
Further, when processing these data, a technique for converting Motion JPEG data into MPEG data is also required.
[0007]
[Problems to be solved by the invention]
However, in the above-described example, when converting Motion JPEG data to MPEG data, the Motion JPEG data is decoded and returned to image data, and the decoded image data is re-encoded into MPEG data. There were some disadvantages.
[0008]
That is, since the JPEG decoding and the MPEG encoding are performed at the same time, the control of the apparatus is complicated, and the processing amount increases. For example, if a JPEG decoder or an MPEG encoding unit is realized by software, it takes a very long time to perform processing. In addition, there is a problem that a work frame memory for storing a reference frame is required when encoding in the MPEG system, which becomes an obstacle to reducing the cost of the entire apparatus.
[0009]
Therefore, an object of the present invention is to reduce intra-frame encoded data to be decoded into image data when converting intra-frame encoded data such as JPEG data into encoded data based on inter-frame correlation such as MPEG data. Another object of the present invention is to provide an image processing method capable of converting data efficiently.
[0010]
[Means for Solving the Problems]
In order to achieve the above object, an image processing method of the present invention includes an input step of inputting first intra-frame encoded data that is intra-frame encoded by a first encoding method for each block; A macroblock in which a plurality of blocks are collected in second intraframe encoded data in a second encoding scheme different from the first encoding scheme by a lookup table without decoding intraframe encoded data. A first data conversion step of converting in units, the first intra-frame encoded data is decoded, and the decoded image data is converted into the second data of Converted into inter-frame encoded data based on the inter-frame correlation of the encoding method for each macroblock. First 2 data conversion step, the first orthogonal transform coefficient data obtained by decoding the first intra-frame encoded data, and the second orthogonal transform coefficient data of the previous frame or the subsequent frame, A comparison step for comparing in units of macroblocks and the number of macroblocks to be skipped before the second intra-frame encoded data and the second intra-frame encoded data according to the comparison result in the comparison step The first to output the macroblock address increment code processing And outputting a macroblock address increment code indicating the number of macroblocks to be skipped before the interframe encoded data and the interframe encoded data. processing And a third that does not output the encoded data. processing In the selection step, when the data difference is larger than a first predetermined value according to the comparison result in the comparison step, the first step processing And when the data difference is less than the first predetermined value and greater than a second predetermined value, the second processing And when the data difference is less than or equal to the second predetermined value, the third processing It is characterized by selecting.
[0022]
【Example】
Hereinafter, an image encoded data conversion apparatus according to a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an image encoded data conversion apparatus according to an embodiment of the present invention.
[0023]
In the following description, intra-frame encoded data is assumed to be Motion JPEG data encoded by the JPEG method, and encoded data using inter-frame correlation is assumed to be data encoded by the MPEG method.
[0024]
In FIG. 1, reference numeral 1 denotes a storage device that stores Motion JPEG data, and reference numeral 2 denotes a JPEG decoder that performs JPEG decoding. Reference numeral 3 denotes a frame memory for storing frame images of moving images. Reference numeral 4 denotes a display for displaying image data stored in the frame memory 3. A data conversion unit 5 converts Motion JPEG data into data encoded by the MPEG method. A communication interface 6 controls communication with the outside of the apparatus. Reference numeral 7 denotes a communication line constituted by a computer network or a public line (for example, an ISDN line). A storage device 8 stores MPEG data encoded by the MPEG method. A CPU 9 controls and controls the entire apparatus. An input device 10 is used by a user to request or specify a moving image display.
[0025]
Conversion of JPEG data to MPEG data is performed according to the following procedure. That is, Motion JPEG data is input from the storage device 1 to the data conversion unit 5. The data converter 5 converts the input Motion JPEG data into MPEG data.
[0026]
FIG. 2 shows a block diagram of the data converter 5. In the figure, reference numeral 101 denotes a code buffer for storing Motion JPEG data for an MPEG encoding unit. Here, an MPEG encoding unit is an MPEG macroblock. The macro block consists of four luminance blocks Y adjacent to each other on the left and right and top and bottom ₀ , Y ₁ , Y ₂ , Y _Three And two chrominance blocks Cb and Cr corresponding to the same position on the image are composed of six blocks. A code memory 102 stores Motion JPEG data for one frame. Reference numeral 103 denotes a decoder for inputting Motion JPEG data and decoding DCT coefficients. Reference numerals 104, 105, and 106 denote latches that store quantization coefficients. Reference numerals 107 and 108 denote latches for storing the decoded DCT coefficients for one macro block of the MPEG system.
[0027]
Reference numeral 109 denotes a difference comparator that calculates an absolute value difference of DCT coefficients in units of blocks, compares each difference value with a preset threshold value, and outputs a determination result. Reference numeral 110 denotes an encoder that generates MPEG data from decoded data. A data converter 111 converts encoded data of Motion JPEG data into MPEG data. Reference numeral 112 denotes an encoder that generates MPEG data based on the output of the difference comparator 109. Reference numeral 113 denotes an encoder that generates MPEG header data. Reference numeral 114 denotes a selector that selects and outputs an input based on the output result of the difference comparator 109. A code buffer 115 stores the output of the selector 114. Reference numeral 116 denotes a header analyzer that analyzes each header data of the Motion JPEG data. Reference numeral 117 denotes a latch that stores the output of the differential output unit 109 for one macroblock.
[0028]
Here, for ease of explanation, it is assumed that the input Motion JPEG data has a sub-sampling ratio of 4: 1: 1. In addition, the structure of an MPEG GOP (Group of Picture) includes an intra-frame coding (Intra) frame (hereinafter abbreviated as I frame) and a motion-compensated coding (Predictive) frame (hereinafter referred to as P (Abbreviated as “frame”).
[0029]
The configuration of an MPEG GOP is composed of I frames and P frames, the configuration of 1 GOP is 15 frames, and the encoding mode of each frame is
I, P, P, P, P, P, P, P, P, P, P, P, P, P, P
And
[0030]
Hereinafter, the operation of each component in FIG. 2 will be described in detail.
First, prior to conversion from Motion JPEG data to MPEG data, the CPU 9 in FIG. 1 initializes each component shown in FIG.
[0031]
Next, the header data of the first frame of the Motion JPEG data is input to the header analyzer 116 via the code buffer 101. The header analyzer 116 analyzes the header data of the Motion JPEG data and obtains image attributes such as an image size. This result is input to the encoder 113. The encoder 113 receives the attributes as a moving image, the frame rate, the encoding rate, and the like from the CPU 9 together with these attributes, and generates MPEG header data based on these. The generated MPEG header data is output to the code buffer 115 via the selector 114. The MPEG header data generated here includes sequence header data starting with a sequence start code (SSC), GOP header data starting with a GOP start code (GSC), and the like. In addition, the header analyzer 116 obtains the quantization coefficient Q1 using a predetermined default quantization table from the quantization table used in the JPEG method. The obtained quantization coefficient Q1 is input to the latch 105. The default quantization table is encoded and placed in the sequence header data.
[0032]
The conversion from Motion JPEG data to MPEG data of I frame will be described below. Prior to intra-frame coding, the CPU 9 stores the quantization coefficient Q2 in the latch 106 for coding control in the MPEG system. Further, the encoder 113 generates frame header data (picture header). Here, it is shown that the encoded frame is an I frame.
[0033]
Motion JPEG data for an MPEG encoding unit is sequentially read from the storage device 1 of FIG. 1 and input to the code buffer 101. The positional relationship in the frame is the same between the JPEG encoding macroblock and the MPEG encoding macroblock. Therefore, one macro block includes 4 luminance blocks and 2 color difference blocks. The Motion JPEG data stored in the code buffer 101 is input to the code memory 102 and the data converter 111. The code memory 102 is composed of a FIFO memory or the like.
[0034]
Subsequently, the macro block data is processed. If the input macroblock data is the left end of the frame, the encoder 113 generates a slice header, outputs it to the code buffer 115 via the selector 114, and stores it. When generating a slice header, the quantization coefficient Q2 is read from the latch 106, encoded, and placed in the slice header.
[0035]
A block diagram of the data converter 111 is shown in FIG. An arithmetic unit 150 receives the quantization coefficients Q1 and Q2 and obtains the quantization multiple P1. Reference numeral 151 denotes a selector that changes the output destination of input encoded data in accordance with a selection signal. Reference numeral 152 denotes a macroblock header generator composed of a ROM or the like. Reference numeral 153 denotes a DC converter configured to convert the encoded data of the DC component of the DCT coefficient, which is constituted by a ROM or the like having the encoded data as an address. Reference numeral 154 denotes an AC converter configured to convert the encoded data of the AC component of the DCT coefficient, which is composed of a ROM or the like having the encoded data as an address. Reference numeral 155 denotes a selector that changes the input destination in accordance with the selection signal.
[0036]
First, the data converter 111 outputs macroblock header data. This header data is header data indicating that a macroblock to be encoded is encoded by intraframe encoding, that is, Intra mode. The CPU 9 causes the input of the selector 155 to select the output of the macroblock header generator 152, reads the macroblock header data from the macroblock header generator 152, and outputs it.
[0037]
Subsequently, the data converter 111 inputs the quantization coefficients Q1 and Q2 from the latch 105 and the latch 106 in addition to inputting the JPEG data in block units. First, a quantization multiple P1 is obtained by the arithmetic unit 150 from the quantization coefficient Q1 input from the latch 105 and the quantization coefficient Q2 input from the latch 106 according to the following equation.
P1 = Q2 / Q1 (1)
[0038]
Subsequently, each block is processed.
The block unit encoded data in JPEG data is composed of a DCCM DC component DPCM result, an AC component quantized result, and a 0-run length Huffman code.
[0039]
First, the conversion of the DC component code will be described. In the JPEG system and MPEG system with a sub-sampling ratio of 4: 1: 1, DPCM is applied in the same order in a Z-shape. Therefore, the DC component is converted using the quantization multiple P1. For example, when the difference value D1 of the input luminance component is 33, the Huffman code is “1110100001”. If the quantization multiple P1 is 3, the output difference value D2 is
D2 = D1 / P1 (2)
Is required.
[0040]
In this case, the difference value D2 is 11, and thus the Huffman code is converted to “1011011”. Actually, the conversion result is stored in advance in the DC converter 153 using the value of the quantization multiple P1 and the Huffman code as an address, and the conversion is performed by reading from the ROM of the DC converter 153.
[0041]
Next, conversion of the AC component code will be described. For example, the luminance component Huffman code “111111111000100111000” indicates that the 0 run length is 2 and the quantization result R1 is 16. If the quantization multiple P1 is 3, the output quantization result R2 is
R2 = R1 / P1 (3)
To Huffman code “1111110111101” in which 0 run length is 2 and quantization result is 5.
[0042]
Actually, the conversion result is stored in advance in the AC converter 154 using the value of the quantization multiple P1 and the Huffman code as addresses, and the conversion is performed by reading from the ROM as the LUT of the AC converter 154.
The MPEG data obtained by such conversion is stored in the code buffer 115.
[0043]
Thereafter, processing is performed in units of macroblocks, and when the processing of one frame is completed, the contents of the code buffer 115 are transmitted and stored in the storage device 8 of FIG. 1 or output to the communication line 7 via the communication interface 6. Or The transmitted code length is reported to the CPU 9. Thereafter, the code buffer 115 is cleared.
[0044]
As described above, according to the present embodiment, when converting Motion JPEG data to MPEG data of I frame, the Motion JPEG data and the quantization coefficient can be used without decoding the Motion JPEG data and returning it to the image data as in the prior art. Since the data is directly converted into MPEG data by primary conversion, the conversion time can be shortened.
[0045]
Next, conversion from Motion JPEG data to MPEG data of P frames will be described.
Prior to motion compensation encoding, the CPU 9 obtains a quantization coefficient Q2 for encoding control from the code length applied to the previous frame and stores it in the latch 106. Further, the quantization coefficient Q1 ′ of the latch 105 is moved to the latch 104. Then, a quantization coefficient Q1 is obtained using a predetermined default quantization table from the quantization table used in the JPEG method. The obtained quantization coefficient Q1 is input to the latch 105. The encoder 113 generates a picture header for the frame. Here, it is shown that the encoded frame is a P frame.
[0046]
Motion JPEG data for one macroblock is sequentially read from the storage device 1 and input to the code buffer 101. The JPEG data stored in the code buffer 101 is input to the code memory 102, the decoder 103, and the data converter 111. Then, the JPEG data input to the decoder 103 is decoded, a DC component and an AC component of the DCT coefficient are generated for each block using the quantization coefficient Q 1 of the latch 105, and stored in the latch 107. Subsequently, the JPEG data of the macro block at the same position of the previous frame is read from the code memory 102, the JPEG data is decoded by the decoder 103, and the DCT coefficient DC is applied to each block using the quantization coefficient Q1 'of the latch 104. A component and an AC component are generated and stored in the latch 108.
[0047]
Thereafter, the difference comparator 109 reads the DCT coefficient for each block from the latch 107 and the latch 108, obtains the difference for each DCT coefficient, and adds the absolute value. After obtaining the addition values A1 to A6 of each block, if all the addition values are 0 and the content of the latch 117 is “10” or “00”, the comparison subtractor 109 sets the code “10”. Generate.
[0048]
Otherwise, the addition values A1 to A6 are compared with a predetermined threshold value T1, and for all addition values,
T1> Ai ≧ 0 (i = 1 to 6) (4)
If so, the difference comparator 109 generates a code “00”.
If even one of the conditions of the expression (4) is not satisfied, a code “01” is generated. The generated code is output to the selector 114.
[0049]
At the same time, the encoder 110 generates macroblock header data. That is, a motion vector code is generated with a motion vector of (0, 0). The encoder 110 generates header data indicating that the macroblock to be encoded is encoded in motion compensation encoding, that is, in predictive mode, and inserts a motion vector code.
[0050]
The encoder 110 reads the DCT coefficient of the macroblock to be encoded from the latch 107, reads the DCT coefficient of the macroblock to be encoded from the latch 108 and the macroblock at the same position in the previous frame, and the quantization coefficient from the latch 106. Read Q2. Then, the encoder 110 performs inverse DCT conversion on the DCT coefficient of each block of the macro block of the latch 107 to reproduce a pixel value. Subsequently, the DCT coefficient of each block of the macro block of the latch 108 is subjected to inverse DCT transform to reproduce the pixel value. Thereafter, the difference between the pixel values is obtained. The obtained difference value is subjected to DCT transformation and quantized with a quantization coefficient Q2, and a quantization result is obtained. Regarding the DC component, if the output code of the difference comparator 109 of the immediately preceding macroblock is “00”, the DPCM is subsequently performed, and if not, the DPCM is newly initialized. A code is assigned to the quantization result of the AC component in accordance with the MPEG system.
[0052]
If the output code of the difference comparator 109 is “00” and the content of the latch 117 is “10”, the selector 114 outputs the content of the encoder 112 and then inputs the output of the encoder 110. The code buffer 115 outputs the data and stores it.
[0053]
If the output code of the difference comparator 109 is “01” and the content of the latch 117 is “10”, the selector 114 outputs the content of the encoder 112, and then the selector 114 outputs the data converter 111. The output is input, output to the code buffer 115, and stored.
[0054]
When the output code of the difference comparator 109 is “10”, the result is input to the encoder 112. When this code is generated, MPEG encoded data is not generated, and the encoder 112 counts the number of consecutively generated codes “10”, and expresses the number (macroblock address increment code). Is generated.
[0055]
When the macroblock processing is completed, the contents of the latch 107 are replaced with the output of the difference comparator 109.
[0056]
Thereafter, processing is performed in units of macroblocks, and when the processing of one frame is completed, the contents of the code buffer 115 are transmitted and stored in the storage device 8 of FIG. 1 or output to the communication line 7 via the communication interface 6. Or The transmitted code length is reported to the CPU 9. Thereafter, the code buffer 115 is cleared.
[0057]
Thereafter, each frame is encoded as an I frame or a P frame in a predetermined order to obtain a desired number of MPEG data. After that, an end code such as a sequence end code determined by the MPEG system is generated by the encoder 113 and sent through the code buffer 115 via the selector 114, and all operations are completed.
[0058]
As described above, according to the present embodiment, when a moving image encoded only by intraframe encoding is converted to encoded data using motion compensation encoding, the circuit scale is large and processing takes time. There is an effect that the encoded data can be instantaneously converted without performing a vector search.
[0059]
In addition, there is no need for a working frame memory for storing reference frames when encoding in the MPEG system, which has the effect of reducing the overall cost of the apparatus.
[0060]
Furthermore, since the orthogonal transformation which takes time for processing can be omitted, there is an effect that processing time can be shortened. Also, high-speed conversion is possible by converting encoded data using an LUT.
[0061]
In this embodiment, the MPEG encoding method has been described as an example. Even in the H.261 encoding system, the configuration of this embodiment is the same as that of H.264. The present invention can be applied by substituting one suitable for the code of H.261.
[0062]
Next, an image encoded data conversion apparatus according to a second embodiment of the present invention will be described with reference to the drawings.
FIG. 4 is a block diagram of the data conversion unit 5 shown in FIG. 1 when converting Motion JPEG data into MPEG data into which a bi-directional motion compensation encoding (Bidirectionaj Predictive) frame (hereinafter abbreviated as B frame) is introduced.
[0063]
The data converter 5 converts the input Motion JPEG data into MPEG data. That is, in FIG. 4, reference numeral 201 denotes a code buffer that stores Motion JPEG data for one macroblock of the MPEG system. Reference numerals 202 and 203 denote code memories for storing Motion JPEG data for one frame. Reference numerals 204, 205, and 206 denote decoders that input Motion JPEG data and decode it into DCT coefficients. A header analyzer 207 analyzes each header data of the Motion JPEG data. 208, 209, 210, and 211 are latches for storing quantization coefficients. Reference numerals 212, 213, and 214 denote latches that store the decoded DCT coefficients for the macroblock.
[0064]
Reference numeral 215 denotes a difference comparator that calculates an absolute value difference of DCT coefficients in units of blocks, compares each difference value with a preset threshold value, and outputs a determination result. A latch 216 stores the output of the difference comparator 215 for one macroblock. Reference numerals 217, 218, and 219 denote encoders that generate MPEG data from the decoded data. A data converter 220 converts encoded data of JPEG data into MPEG data. Reference numeral 221 denotes an encoder that generates MPEG header data. Reference numeral 222 denotes an encoder that generates MPEG data based on the output of the difference comparator 215. Reference numeral 223 denotes a selector that selects and outputs an input based on the output result of the difference comparator 215. A code buffer 224 stores the output of the selector 223.
[0065]
Here, for ease of explanation, it is assumed that the input Motion JPEG data has a sub-sampling ratio of 4: 1: 1. Also, the MPEG GOP configuration is composed of 1 frame, P frame, and B frame, 1 GOP configuration is 15 frames, and the encoding mode of each frame is
I, B, B, P, B, B, P, B, B, P, B, B, P, B, B
And
[0066]
Hereinafter, the operation of each component in FIG. 4 will be described in detail.
First, prior to conversion from Motion JPEG data to MPEG data, the CPU 9 initializes each component shown in FIG.
[0067]
Next, the header data of the first frame of the Motion JPEG data is input to the header analyzer 207 via the code buffer 201. The header analyzer 207 analyzes the header data of the Motion JPEG data and obtains image attributes such as an image size. This result is input to the encoder 221. The encoder 221 receives the attributes of the moving image, the frame rate, the encoding rate, and the like from the CPU 9 together with these attributes, generates MPEG header data, and outputs the MPEG header data to the code buffer 224 via the selector 223. The header data generated here is a sequence header starting with a sequence start code and GOP header data starting with a GOP start code. Also, the quantization coefficient Q3 is obtained using a predetermined default quantization table from the quantization table used in the JPEG method. The obtained quantization coefficient Q3 is input to the latch 208. The default quantization table is encoded and put into sequence header data.
[0068]
The conversion from Motion JPEG data to MPEG data of I frame will be described below. Prior to intra-frame coding, the CPU 9 stores the quantization coefficient Q4 in the latch 211 for code control. Also, the encoder 221 generates a picture header. Here, it is shown that the encoded frame is an I frame.
[0069]
First, Motion JPEG data for one macroblock is sequentially read from the storage device 1 and input to the code buffer 201. The JPEG data stored in the code buffer 201 is input to the code memory 202 and the data converter 220. The input JPEG data is stored in the code memory 202. Similarly, the Motion JPEG data of the macro block at the same position as input from the code memory 202 is stored in the code memory 203. The code memories 202 and 203 are composed of FIFO memories. The configuration of the data converter 220 is configured by an LUT, similar to the data converter 111 of the first embodiment.
[0070]
Subsequently, the macro block data is processed. If the input macroblock data is the left end of the frame, the encoder 221 generates a slice header, outputs it to the code buffer 224 via the selector 223, and stores it. Further, when generating a slice header, the quantization coefficient Q4 is read from the latch 211, encoded, and placed in the slice header.
[0071]
The data converter 220 generates macroblock header data. That is, the data converter 220 generates header data indicating a macroblock encoded in the Intra mode.
[0072]
The data converter 220 inputs the quantization coefficients Q3 and Q4 from the latch 208 and the latch 211 in addition to inputting the JPEG data in units of blocks. First, a quantization multiple P2 is obtained from the quantization coefficient Q3 input from the latch 208 and the quantization coefficient Q4 input from the latch 211 according to the following equation.
P2 = Q4 / Q3 (5)
[0073]
Subsequently, each block is processed.
First, the conversion of the DC component code will be described. In this case, the DC component is converted using the quantization multiple P2. For example, if the difference value D3 of the input luminance component is 33, if the quantization multiple P is 1.5, the output difference value D4 is
[0074]
D4 = D3 / P2 (6)
The difference value D4 in this case is 22. Therefore, this Huffman code is converted to “11010110”.
[0075]
Next, conversion of the AC component code will be described. For example, the luminance component Huffman code “111111111000100111000” indicates that the 0 run length is 2 and the quantization result R3 is 16. If the quantization multiple P2 is 1.5, the output quantization result R4 is
R4 = R3 / P2 (7)
To 10.6, and is rounded off to be converted into a Huffman code “1111111101001011” in which the 0 run length is 2 and the quantization result is 11.
[0076]
These conversions are performed by referring to a LUT (Look Up Table) composed of a ROM or the like.
The MPEG data obtained by such conversion is stored in the code buffer 224.
[0077]
Thereafter, processing is performed in units of macroblocks, and when processing of one frame is completed, the contents of the code buffer 224 are transmitted and stored in the storage device 8 of FIG. 1 or output to the communication line 7 via the communication interface 6. Or The transmitted code length is reported to the CPU 9. Thereafter, the code buffer 224 is cleared.
[0078]
Next, conversion from Motion JPEG data to MPEG data of P frames will be described.
Prior to encoding the P frame, the CPU 9 obtains a quantization coefficient Q4 for encoding control from the code length applied to the previous frame and stores it in the latch 211. Further, the quantization coefficient Q3 ″ of the latch 209 is moved to the latch 210, and the quantization coefficient Q3 ′ of the latch 208 is moved to the latch 209. Then, a quantization coefficient Q3 is obtained using a predetermined default quantization table from the quantization table used in the JPEG method. The obtained quantization coefficient Q3 is input to the latch 208. The encoder 221 generates a frame picture header. Here, it is shown that the encoded frame is a P frame.
[0079]
Motion JPEG data for one macroblock is sequentially read from the storage device 1 and input to the code buffer 201. The JPEG data stored in the code buffer 201 is input to the code memory 202, the decoder 204, and the data converter 220. The decoder 204 decodes the input JPEG data, generates a DC component and an AC component of the DCT coefficient for each block using the quantization coefficient Q 3 of the latch 208, and stores them in the latch 212. At the same time, the JPEG data of the macroblock at the same position of the previous I frame or P frame is read from the code memory 202, decoded by the decoder 205, and the quantization coefficient Q3 'of the latch 209 is used to determine the DCT coefficient for each block. A DC component and an AC component are generated and stored in the latch 213. The read JPEG data is also input to the code memory 203.
[0080]
After that, the difference comparator 215 reads the DCT coefficient for each block from the latch 212 and the latch 213, calculates the difference for each DCT coefficient, and adds the absolute value. After obtaining the addition values A1 to A6 of each block, if all the addition values are 0 and the contents of the latch 216 are “110” or “000”, the comparison subtractor 215 sets the code “110”. Generate.
[0081]
In other cases, the addition values A1 to A6 are compared with a predetermined threshold value T2, and for all addition values,
T2> Ai ≧ 0 (i = 1 to 6) (8)
If so, the difference comparator 215 generates a code “000”.
[0082]
If even one of the conditions of equation (8) is not satisfied, code “100” is generated. The generated code is output to the selector 223.
[0083]
In parallel with this, the encoder 217 and the data converter 220 perform the following operations. The encoder 221 generates macroblock header data. That is, a motion vector code is generated with a motion vector of (0, 0). The encoder 221 generates header data indicating that the macroblock to be encoded is encoded in the predictive mode, and inserts a motion vector code.
[0084]
The encoder 217 first reads the DCT coefficient of the macroblock to be encoded from the latch 212, and also reads the DCT coefficient of the macroblock at the same position in the previous I frame or P frame from the latch 214, The quantization coefficient Q4 is read from the latch 211. Then, each block DCT coefficient of the macro block of the latch 212 is subjected to inverse DCT transform to reproduce a pixel value. Subsequently, the DCT coefficient of each block of the macro block of the latch 214 is subjected to inverse DCT transform to reproduce the pixel value. Thereafter, a difference between the respective pixel values is obtained, DCT transformation is performed on the obtained difference value, and quantization is performed with a quantization coefficient Q4 to obtain a quantization result. Regarding the DC component, if the output code of the difference comparator 215 of the immediately preceding macroblock is “000”, the DPCM is subsequently performed. Otherwise, the DPCM is newly initialized. A code is assigned to the quantization result of the AC component in accordance with the MPEG system.
[0085]
The JPEG data input to the data changer 220 is converted from JPEG data to MPEG data in the same manner as the Intra frame macroblock.
[0086]
If the output code of the difference comparator 215 is “000” and the content of the latch 216 is “110”, the selector 223 outputs the content of the encoder 222 and then inputs the output of the encoder 218. Then, it outputs to the code buffer 224 and stores it.
[0087]
If the output code of the difference comparator 215 is “100” and the content of the latch 216 is “110”, the selector 223 outputs the content of the encoder 222, and then the selector 223 outputs the data converter 220. The output is input, output to the code buffer 224, and stored.
[0088]
When the output code of the difference comparator 215 is “110”, the result is input to the encoder 222. When this code is generated, MPEG encoded data is not generated, and the encoder 222 counts the number of consecutively generated codes “110” to generate a macroblock address increment code.
[0089]
When the macroblock processing is completed, the contents of the latch 216 are replaced with the output of the difference comparator 215.
[0090]
Thereafter, processing is performed in units of macroblocks, and when processing of one frame is completed, the contents of the code buffer 224 are transmitted and stored in the storage device 8 of FIG. 1 or output to the communication line 7 via the communication interface 6. Or The transmitted code length is reported to the CPU 9. Thereafter, the code buffer 224 is cleared.
[0091]
Finally, conversion from Motion JPEG data to B frame MPEG data will be described.
Prior to encoding the B frame, the CPU 9 obtains a quantization coefficient Q4 for encoding control from the code length applied to the previous frame and stores it in the latch 211. Further, the quantization coefficient Q3 ″ of the latch 209 is moved to the latch 210, and the quantization coefficient Q3 ′ of the latch 208 is moved to the latch 209. Then, a quantization coefficient Q3 is obtained using a predetermined default quantization table from the quantization table used in the JPEG method. The obtained quantization coefficient Q3 is input to the latch 208. The encoder 221 generates a frame picture header. Here, it is shown that the encoded frame is a B frame.
[0092]
Motion JPEG data for one macroblock is sequentially read from the storage device 1 and input to the code buffer 201. The JPEG data stored in the code buffer 201 is input to the decoder 204 and the data converter 220. The decoder 204 decodes the input JPEG data, generates a DC component and an AC component of the DCT coefficient for each block using the quantization coefficient Q 3 of the latch 208, and stores them in the latch 212. At the same time, the JPEG data of the macro block at the same position of the subsequent I frame or P frame is read from the code memory 202, and the JPEG data of the macro block at the same position of the previous I frame or P frame is read from the code memory 203. The JPEG data read from the code memory 202 is decoded by the decoder 205, DCT DC components and AC components are generated for each block using the quantization coefficient Q 3 ′ of the latch 209, and stored in the latch 213. . The JPEG data read from the code memory 203 is decoded by the decoder 206, generates DCT DC components and AC components for each block using the quantization coefficient Q 3 ″ of the latch 210, and is stored in the latch 214. To do.
[0093]
Thereafter, the difference comparator 215 reads the DCT coefficient for each block from the latches 212, 213, and 214, obtains the difference for each DCT coefficient, and adds the absolute value. That is, the difference comparator 215 obtains the absolute value addition value BA1 to BA6 of the difference for each DCT coefficient of each block of the latch 212 and the latch 213. Further, FA6 is obtained from the added value FA1 of the absolute value of the difference for each DCT coefficient of each block of the latch 212 and the latch 214. Further, an average value for each DCT coefficient of each block of the latches 213 and 214 is obtained, and an addition value MA1 to MA6 of absolute values of differences from the respective coefficients of the latch 212 is obtained.
[0094]
If all the addition values FA1 to FA6 are 0, and the content of the latch 216 is “101” or “000”, the comparison subtractor 215 generates a code “101”.
If all the addition values BA1 to BA6 are 0 and the content of the latch 216 is “110” or “001”, the comparison subtractor 215 generates the code “110”.
If all the addition values MA1 to MA6 are 0 and the content of the latch 216 is “111” or “010”, the comparison subtractor 215 generates the code “111”.
[0095]
In other cases, the minimum value SA is selected from the addition value FA to which all the addition values FA1 to 6 are added, the addition value BA to which all the addition values BA1 to 6 are added, and the addition value MA to which all the addition values MA1 to 6 are added. select. That is, the minimum value SA is
SA = min (FA, BA, MA) (9)
It is expressed.
[0096]
After that, the minimum value SA is compared with a predetermined threshold value T5, and the minimum value SA is
T5> SA (10)
If the added value FA is minimum, the difference comparator 215 generates a code “000”. If the addition value BA is minimum, the difference comparator 215 generates a code “001”. If the added value MA is minimum, the difference comparator 215 generates a code “010”.
[0097]
Further, when none of the conditions of the expression (10) is satisfied, the difference comparator 215 generates a code “011”. The generated code is output to the selector 223.
[0098]
At the same time, the encoder 217, the encoder 218, the encoder 219, the data converter 220, and the encoder 221 perform the following operations.
[0099]
First, the encoder 221 generates macroblock header data. That is, a motion vector code is generated with the motion vector from the previous frame in time as (0, 0). At this time, the encoder 221 generates header data indicating that the macroblock to be encoded is encoded in the Forward Predictive mode, and inserts a motion vector code.
[0100]
The encoder 217 first converts the DCT coefficient of the macroblock to be encoded from the latch 212 into the DCT coefficient of the macroblock at the same position in the I frame or P frame temporally prior to the macroblock to be encoded from the latch 214. The quantization coefficient Q4 is read from the latch 211. The encoder 217 further reproduces a pixel value by performing inverse DCT conversion on the DCT coefficient of each block of the macro block of the latch 212. Subsequently, the DCT coefficient of each block of the macro block of the latch 214 is subjected to inverse DCT transform to reproduce the pixel value. Thereafter, a difference between the respective pixel values is obtained, DCT transformation is performed on the obtained difference value, and quantization is performed with a quantization coefficient Q4 to obtain a quantization result. Regarding the DC component, if the output code of the difference comparator 215 of the immediately preceding macroblock is “000”, then the DPCM is performed, otherwise the DPCM is newly initialized. A code is assigned to the quantization result of the AC component in accordance with the MPEG system.
[0101]
Also, the encoder 221 generates macroblock header data. That is, a motion vector code is generated with a motion vector from a temporally subsequent frame as (0, 0). At this time, the encoder 221 generates header data indicating that the macroblock to be encoded is encoded in the Backword Predictive mode, and inserts a motion vector code.
[0102]
The encoder 218 first reads the DCT coefficient of the macroblock to be encoded from the latch 212, and the macroblock at the same position in the I frame or the P frame later in time as the macroblock to be encoded from the latch 213. The DCT coefficient is read, and the quantization coefficient Q4 is read from the latch 211. The encoder 218 further performs inverse DCT conversion on the DCT coefficient of each block of the macro block of the latch 212 to reproduce a pixel value. Subsequently, the DCT coefficient of each block of the macro block of the latch 213 is subjected to inverse DCT transform to reproduce the pixel value. Thereafter, a difference between the respective pixel values is obtained, DCT transformation is performed on the obtained difference value, and quantization is performed with a quantization coefficient Q4 to obtain a quantization result. Regarding the DC component, if the output code of the difference comparator 215 of the immediately preceding macroblock is “001”, then the DPCM is performed, otherwise the DPCM is newly initialized. A code is assigned to the quantization result of the AC component in accordance with the MPEG system.
[0103]
Furthermore, the encoder 221 generates macroblock header data. That is, the motion vector code is generated with each motion vector from the previous and subsequent frames as (0, 0). At this time, the encoder 221 generates header data indicating that the macroblock to be encoded is encoded in the Bidirectional Predictive mode, and inserts both motion vector codes.
[0104]
The encoder 219 first reads the DCT coefficient of the macroblock to be encoded from the latch 212, and the macroblock at the same position in the I frame or P frame that is temporally later from the macroblock to be encoded from the latch 213. The DCT coefficient is read, the DCT coefficient of the macroblock at the same position in the previous I frame or P frame as the macroblock to be encoded is read from the latch 214, and the quantization coefficient Q4 is read from the latch 211. The encoder 219 further performs inverse DCT conversion on the DCT coefficient of each block of the macroblock of the latch 213 to reproduce a pixel value, and inverse DCT conversion of the DCT coefficient of each block of the macroblock of the latch 214. Reproduce. An average is obtained for each pixel value. Subsequently, the DCT coefficient of each block of the macro block of the latch 212 is subjected to inverse DCT transform to reproduce the pixel value. Thereafter, the difference between the pixel value and the average value is obtained for each pixel, the obtained difference value is subjected to DCT transform, and quantized with the quantization coefficient Q4 to obtain a quantization result. Regarding the DC component, if the output code of the difference comparator 215 of the immediately preceding macroblock is “010”, the DPCM is subsequently performed, otherwise the DPCM is newly initialized. A code is assigned to the quantization result of the AC component in accordance with the MPEG system.
[0105]
The JPEG data input to the data converter 220 is converted from JPEG data to MPEG data in the same manner as an intra frame macroblock.
[0106]
When the output code of the difference comparator 215 is “000”, if the content of the latch 216 is “101”, “110”, or “111”, the selector 223 outputs the content of the encoder 222, The output of the encoder 217 is input, output to the code buffer 224, and stored.
[0107]
When the output code of the difference comparator 215 is “001”, if the content of the latch 216 is “101”, “110”, or “111”, the selector 223 outputs the content of the encoder 222, The output of the encoder 218 is input, output to the code buffer 224, and stored.
[0108]
If the output code of the difference comparator 215 is “010” and the content of the latch 216 is “101”, “110” or “111”, the selector 223 outputs the content of the encoder 222, The output of the encoder 219 is input, output to the code buffer 224, and stored.
[0109]
When the output code of the difference comparator 215 is “011”, if the content of the latch 216 is “101”, “110”, or “111”, the selector 223 outputs the content of the encoder 222, The selector 223 receives the output of the data converter 220, outputs it to the code buffer 224, and stores it.
[0110]
When the output code of the difference comparator 215 is “101”, the result is input to the encoder 222. When this code is generated, MPEG encoded data is not generated, and the encoder 222 counts the number of consecutively generated codes “101” to generate a macroblock address increment code.
[0111]
When the output code of the difference comparator 215 is “110”, the result is input to the encoder 222. When this code is generated, MPEG encoded data is not generated, and the encoder 222 counts the number of consecutively generated codes “110” to generate a macroblock address increment code.
[0112]
When the output code of the difference comparator 215 is “111”, the result is input to the encoder 222. When this code is generated, MPEG encoded data is not generated, and the encoder 222 counts the number of consecutively generated codes “111” to generate a macroblock address increment code.
[0113]
When the macroblock processing is completed, the contents of the latch 216 are replaced with the output of the difference comparator 215.
[0114]
Thereafter, processing is performed in units of macroblocks, and when processing of one frame is completed, the contents of the code buffer 224 are transmitted and stored in the storage device 8 of FIG. 1 or output to the communication line 7 via the communication interface 6 To do. The transmitted code length is reported to the CPU 9. Thereafter, the code buffer 224 is cleared.
[0115]
Thereafter, each frame is encoded as an I frame or a P frame in a predetermined order to obtain a desired number of MPEG data. Thereafter, an end code such as a sequence end code determined by the MPEG method is generated by the encoder 221 and sent via the selector 223 via the code buffer 224, and all operations are completed.
[0116]
Next, an image encoded data conversion apparatus according to a third embodiment of the present invention will be described with reference to FIG.
The third embodiment of the present invention shows a case where data processing is performed using software in an arithmetic device, and FIG. 5 shows the configuration of an image encoded data conversion device according to the third embodiment of the present invention. FIG. 6 to FIG. 14 are flowcharts showing the procedure of image encoded data conversion according to the third embodiment of the present invention. In FIG. 5, the same components as those in FIG.
[0117]
In FIG. 5, reference numeral 301 denotes a CPU that controls the entire apparatus and converts Motion JPEG data into MPEG data. Reference numeral 302 denotes a program memory for storing software for data processing, and reference numeral 303 denotes a work memory for securing a work area for processing.
[0118]
For ease of explanation, it is assumed that the input Motion JPEG data has a sub-sampling ratio of 4: 1: 1. The MPEG GOP is composed of an I frame and a P frame.
[0119]
First, as shown in step S01 of FIG. 6, the CPU 301 reads a program for converting Motion JPEG data / MPEG data from the program memory 302 and starts execution.
[0120]
Next, as shown in step S <b> 02, a conversion result LUT (DC component conversion LUT) having the quantization multiple P <b> 3 and the DC component Huffman code as addresses is stored in the work memory 303.
[0121]
Next, as shown in step S <b> 03, a conversion result LUT (AC component conversion LUT) having the value of the quantization multiple P <b> 3 and the AC component Huffman code as addresses is stored in the work memory 303.
[0122]
Next, as shown in step S04, the header data of the frame of Motion JPEG data is read, the header data of JPEG data is analyzed, and image attributes such as image size are obtained.
[0123]
Next, as shown in step S05, the attribute as a moving image is added together with each attribute and default quantization table, and a sequence header and GOP header data of MPEG data are generated and output.
[0124]
Hereinafter, I frame conversion will be described.
First, as shown in step S09, it is determined whether the frame to be generated is an I frame or a P frame. If an I frame is to be generated, the process proceeds to step S10.
[0125]
Next, as shown in step S10, a quantization coefficient Q8 is obtained from the quantization table of JPEG encoded data using the JPEG default quantization table.
[0126]
Next, as shown in step S11, a quantization coefficient Q9 is determined for encoding control in the MPEG system.
[0127]
Next, as shown in step S12 of FIG. 7, a picture header indicating an I frame is generated.
[0128]
Next, as shown in step S <b> 14, Motion JPEG data for one macroblock is sequentially read from the storage device 1. The read Motion JPEG data is stored in the work memory 303 in units of macro blocks.
[0129]
If the input macroblock data is the left end of the frame as shown in step S15, the quantization coefficient Q9 is read out as shown in step S16, and a slice header is generated and output.
[0130]
Next, as shown in step S <b> 17, prior to processing the macroblock, macroblock header data representing encoding in the intra mode is generated and output.
[0131]
Next, as shown in step S18, JPEG data is processed in units of blocks. The quantization multiple P3 is obtained from the quantization coefficient Q8 and the quantization coefficient Q9 according to the following equation.
P3 = Q8 / Q9 (11)
[0132]
Next, as shown in step S20 of FIG. 8, conversion from JPEG data in the DC component to MPEG data is performed with reference to the DC component conversion LUT in the working memory 303 from the quantization multiple P3 and the Huffman code. The encoded data obtained by the conversion is output.
[0133]
For example, when the difference value D5 of the JPEG data of the direct current component of the luminance component is 20, the Huffman code is “11010100”. Here, if the quantization multiple P3 is 2.5, the difference value D6 of the output MPEG data is
D6 = D5 / P3 (12)
Is required.
[0134]
In this case, the difference value D6 is 4, and thus the Huffman code is converted to “100100”. The encoded data obtained by the conversion is temporarily stored in the work memory 303.
[0135]
Next, as shown in step S22, the conversion from the JPEG data to the MPEG data in the AC component is similarly performed by referring to the AC component conversion LUT in the working memory 303 from the quantization multiple P3 and the Huffman code. The encoded data obtained by the conversion is output.
[0136]
The quantization result R6 of the MPEG data after conversion from the quantization result R5 of the input JPEG data and the quantization multiple P3 is:
R6 = R5 / P3 (13)
Is required.
[0137]
For example, it is assumed that the Huffman code “1111111011111” indicating that the quantization multiple P3 is 7, the 0 run length is 5, and the quantization result R5 is 3 is input. The quantization result R6 at this time becomes 1 or less, but is rounded up to 1. Therefore, it is converted to a Huffman code “11110101” in which the 0 run length is 5 and the quantization result is 1.
[0138]
The conversion of the AC component as described above is continued until all AC components are completed as shown in Step S21, and when completed, the process returns to Step S19.
[0139]
Hereinafter, as shown in step S19, the processing is performed in units of macroblocks, and when the processing of one frame is completed, the processing moves to the next frame as shown in step S13 of FIG.
[0140]
In FIG. 5, if there is no next frame as shown in step S06, the moving picture has ended, so that a sequence end code is generated and output as shown in step S07, and the process goes to step S08. As shown, the process ends.
[0141]
Next, conversion of P frame data will be described.
First, as shown in step S09, it is determined whether the frame to be generated is an I frame or a P frame. If a P frame is to be generated, the process proceeds to step S23.
[0142]
Next, as shown in step S23, the quantization coefficient Q10 is obtained from the quantization table of JPEG encoded data using the JPEG default quantization table.
[0143]
Next, as shown in step S24, the quantization coefficient of the JPEG data of the previous frame (quantization coefficient Q8 described above if the previous frame is an I frame, quantization coefficient Q10 of the previous frame if the previous frame is P frame) is quantized. This is held as the conversion factor Q11.
[0144]
Next, as shown in step S25 of FIG. 9, a quantization coefficient Q12 is determined for encoding control.
[0145]
Next, as shown in step S26, a picture header indicating a P frame is generated.
[0146]
Next, as shown in step S <b> 28, Motion JPEG data for one macroblock is sequentially read from the storage device 1. The read Motion JPEG data is stored in the work memory 303 in units of macro blocks.
[0147]
If the input macroblock data is the left end of the frame as shown in step S29, the quantized coefficient Q12 is read out and a slice header is generated and output as shown in step S30.
[0148]
Subsequently, JPEG data is processed in macroblock units.
That is, as shown in step S32 of FIG. 10, the JPEG data is decoded, and the DCT DC component and AC component of each block are generated using the quantization coefficient Q10 and the JPEG default quantization table.
[0149]
Next, as shown in step S33, the JPEG data of the macroblock at the same position in the previous frame is read from the working memory 303 and decoded, and the DCT coefficient of each block using the quantization coefficient Q11 and the JPEG default quantization table. The direct current component and alternating current component of are generated.
[0150]
Next, as shown in step S34, a difference for each DCT coefficient is obtained for each block, and an absolute value is added. These processes are performed for all six blocks, and when the process is completed, the process proceeds to step S35. Note that the addition results are AA1 to AA6 in the order of blocks.
[0151]
If all the added values AA1 to AA6 are 0 as shown in step S35, that is,
AAi = 0 (i = 1 to 6) (14)
Is a macro block in which no code is generated with the motion vector (0, 0).
[0152]
In this case, as shown in step S36, if the encoding mode of the immediately preceding macroblock is the Predictive mode, a skip process is performed.
[0153]
That is, skip macroblocks are counted as shown in step S37.
[0154]
If the encoding mode of the immediately preceding macroblock is not Predictive mode as shown in Step S36, it is encoded in the Predictive mode of the motion vector (0, 0) as shown in Step S53 of FIG. A macro block header representing is output.
[0155]
Furthermore, as shown in step S54, the count value of the macroblock to be skipped is set to zero.
[0156]
If the block addition values AA1 to AA6 are not all 0 as shown in step S35 of FIG. 10, the addition values AA1 to AA6 are compared with a predetermined threshold T3 as shown in step S38 of FIG. And for all added values,
T3>AAi> 0 (i = 1 to 6) (15)
If so, a code in Predictive mode is generated.
[0157]
Next, as shown in step S39, macroblock header data indicating that encoding is performed in the predictive mode is generated. In this header data, a code representing the motion vector (0, 0) and a code representing the position of the block whose added values AA1 to AA6 are not 0 are generated and output in the macroblock header.
[0158]
Next, as shown in step S41, the DCT coefficient of the block in which the addition value AAi (i = 1 to 6) of the frame to be encoded is not 0 is subjected to inverse DCT transform to reproduce the pixel value.
[0159]
Next, as shown in step S42, the pixel value is reproduced by performing inverse DCT conversion on the DCT coefficient of the block at the same position of the frame immediately before encoding, in which the addition value AAi (i = 1 to 6) is not 0. To do.
[0160]
Thereafter, as shown in step S43, the difference between the pixel values is obtained.
[0161]
Next, as shown in step S44, the obtained difference value is subjected to DCT transformation and quantized with a quantization coefficient Q12 to obtain a quantization result. The above processing is performed for all the blocks as shown in step S40, and when finished, the process proceeds to step S45.
[0162]
As shown in step S45 of FIG. 12, for the DC component, if the immediately preceding macroblock is encoded in the predictive mode, DPCM is performed as shown in step S46, and encoding is performed using the MPEG code. ,Output. Also, if the mode is the previous macroblock mode Intra mode or if it is the leftmost macroblock, the DPCM is newly initialized as shown in step S47, and the DC component is encoded as shown in step S48. Do.
[0163]
Next, as shown in step S49, the AC component quantization result is assigned a code in accordance with the MPEG system and is output.
[0164]
Next, as shown in step S50, the encoding of the direct current component in step S51 and the encoding of the alternating current component in step S52 are performed for all the blocks, and when finished, the process returns to step S27.
[0165]
Also, as shown in step S38 of FIG. 11, if even one of the conditions of equation (15) is not satisfied, it is converted into an intra mode code and output.
[0166]
If the previous macroblock has been skipped prior to conversion, it is output as a macroblock address increment code corresponding to the count value, as shown in step S55.
[0167]
For details of the conversion, the same processing as the conversion processing of the macroblock of the I frame in steps S56 to S61 in FIG. 14 is performed.
[0168]
Thereafter, processing is performed in units of macroblocks, and when processing for one frame is completed as shown in step S06 of FIG. 6, the processing for the next frame is performed.
[0169]
If there is no next frame, the moving image is completed, so that a sequence end code is generated and output as shown in step S07, and the process is ended as shown in step S08.
[0170]
In this embodiment, the MPEG encoding method has been described as an example. Even in the H.261 encoding system, the configuration of this embodiment is the same as that of H.264. The present invention can be applied by substituting one suitable for the code of H.261. Further, the case where an MPEG GOP is composed of an I frame and a P frame has been described. However, the present invention may be applied to conversion to an MPEG system in which a B frame is introduced. Further, in this embodiment, when performing the macroblock conversion process of the I frame, steps S56 to S61 in FIG. 14 are used. However, the present invention is not limited to this, and other processes may be used.
[0171]
In addition, as an example of moving image encoding, MPEG and H.264 are used. Although the H.261 method is taken as an example, the moving image encoding method is not limited to this, and any encoding that performs motion compensation may be used. Further, the example of the image sampling of 4: 1: 1 has been described as an example, but even when 4: 4: 4 or 4: 2: 2 is input, only the chromaticity component is reproduced and sub-sampled before being encoded. It ’s fine.
[0172]
【The invention's effect】
As described above, according to the present invention, the first intra-frame encoded data that has been intra-frame encoded by the first encoding scheme is converted into the second frame and the inter-frame of the second encoding scheme. The data can be efficiently converted into encoded data, and the amount of data can be greatly reduced.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of an image encoded data conversion apparatus according to an embodiment of the present invention.
FIG. 2 is a block diagram showing a configuration of a data conversion unit according to the first embodiment of the present invention.
FIG. 3 is a block diagram showing a configuration of a data converter according to an embodiment of the present invention.
FIG. 4 is a block diagram illustrating a configuration of a data conversion unit according to a second embodiment of the present invention.
FIG. 5 is a block diagram showing a configuration of an image encoded data conversion apparatus according to a third embodiment of the present invention.
FIG. 6 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 7 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 8 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 9 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 10 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 11 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 12 is a flowchart showing the procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 13 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
FIG. 14 is a flowchart showing a procedure of image encoded data conversion according to the third embodiment of the present invention.
[Explanation of symbols]
1, 8 Storage device
2 JPEG decoder
3 frame memory
4 display
5 Data converter
6 Communication interface
7 Communication line
9, 301 CPU
10 Input device
101, 115, 201, 224 Code buffer
102, 202, 203 Code memory
103, 204, 205, 206 Decoder
104, 105, 106, 107, 108, 117, 208, 209, 210, 211, 212, 213, 214, 216 Latch
109,215 Difference comparator
110, 112, 113, 217, 218, 219, 221, 222 Encoder
111, 220 Data converter
114, 151, 155, 223 selector
116, 207 Header analyzer
150 calculator
152 Macroblock header generator
153 DC converter
154 AC converter
302 program memory
303 Working memory

Claims (1)

An input step of inputting first intra-frame encoded data that is intra-frame encoded by the first encoding method for each block;
The first intra-frame encoded data is decoded into a second intra-frame encoded data in a second encoding scheme different from the first encoding scheme by a lookup table without decoding. A first data conversion step of converting in units of collected macroblocks;
Decoding the first intra-frame coded data, the inter-coded data and the decoded image data by inter-frame correlation of the second encoding method, the second that converts in the macroblock A data conversion process;
Comparison in which the first orthogonal transform coefficient data obtained by decoding the first intra-frame encoded data and the second orthogonal transform coefficient data of the previous frame or the subsequent frame are compared in units of the macroblock. Process,
A first macroblock address increment code indicating the number of macroblocks to be skipped before the second intra-frame encoded data and the second intra-frame encoded data is output according to the comparison result in the comparison step. A second process for outputting the inter-frame encoded data and a macro block address increment code indicating the number of macro blocks to be skipped before the inter-frame encoded data; and a third process for not outputting the encoded data. And a selection step for selecting the processing of
In the selection step, the first process is selected when the data difference is larger than a first predetermined value according to the comparison result in the comparison step. image processing method characterized by selecting the second processing when the value is greater than said data difference selects the third process if: the second predetermined value.

Images