WO2006093306A1

WO2006093306A1 - Image encoding method conversion device and method

Info

Publication number: WO2006093306A1
Application number: PCT/JP2006/304210
Authority: WO
Inventors: Toru Taima
Original assignee: Pioneer Corporation
Priority date: 2005-03-03
Filing date: 2006-02-28
Publication date: 2006-09-08

Abstract

There are provided an image encoding method conversion device and method including: decoding means for generating decoded image data by decoding image data encoded by a first encoding method and detecting an encoding coefficient for each macro block; encoding coefficient conversion means for converting the encoding coefficient of the first encoding method for each macro block detected by the decoding means into an encoding coefficient of a second encoding method; and encoding means for encoding the decoded image data by using the encoding coefficient of the second encoding method converted by the encoding coefficient conversion means for each macro block and generating image data encoded by the second encoding method.

Description

Image coding system conversion apparatus and method

Technical field

The present invention relates to an image encoding method conversion apparatus and method for converting encoded image data into encoded image data of another encoding method.

Background art

There are many image encoding methods for compressing image data, such as MPEG, MPEG 2, H. 26 61, etc., for converting encoded image data into encoded image data of other encoding methods. An image coding system conversion apparatus is used.

As a conventional image encoding method conversion device, an input frame rate μ measuring means for measuring the frame rate of the decoded image data after decoding the encoded image data corresponding to the encoding method, and its decoding A target code amount calculating means for calculating a target code amount for each picture based on the transmission capacity of the line for transmitting the converted image data and the input frame rate measured by the input frame rate measuring means, and dividing the picture picture Calculating a ratio of a predetermined unit code amount obtained in a predetermined unit for each predetermined unit, and determining a target code amount for each predetermined unit based on the calculation result and a target code amount for each picture; Based on the determination, there is an apparatus provided with an encoding means for encoding the decoded image data by another encoding method (see Japanese Patent Laid-Open No. 2 0 0 3- 1 8 9 3 1 1). See)

However, in such a conventional image coding method conversion apparatus, the re-encoding method is used without using the information of the original coding method except for the motion vector of the macro block. Since the target code amount of each picture is determined and re-encoding is simply performed, the encoded image data obtained by re-encoding the decoded image data has the disadvantage of causing image quality degradation. 7

Disclosure of the invention

The problems to be solved by the present invention include the above-mentioned drawbacks as an example. An image coding method conversion apparatus and method capable of preventing image quality deterioration due to re-encoding by different coding methods, and a method thereof are provided. It is an object of the present invention to provide a computer to be executed and a readable program.

The image coding method conversion apparatus of the present invention converts an image data encoded by the first encoding method into image data encoded by a second encoding method different from the first encoding method. An encoding method conversion device, comprising: decoding means for decoding image data encoded by the first encoding method to generate decoded image data and detecting an encoding coefficient for each macroblock; Coding coefficient conversion means for converting the coding coefficient of the first coding method detected by the decoding means into the coding coefficient of the second coding method for each macroblock; and the decoded image A code for generating image data encoded by the second encoding method by encoding data using the encoding coefficient of the second encoding method converted by the encoding coefficient conversion means for each macroblock. And means It is characterized in that was.

The image encoding method conversion method of the present invention is an image for converting image data encoded by the first encoding method into image data encoded by a second encoding method different from the first encoding method. An encoding method conversion method, wherein the image data encoded by the first encoding method is decoded to generate decoded image data, and for each macro block A decoding step for detecting a coding coefficient, and a code for converting the coding coefficient of the first coding method for each macro block detected in the decoding step into a coding coefficient of the second coding method. Encoding the encoded image data using the encoding coefficient of the second encoding method obtained by converting the decoded image data in the encoding coefficient conversion step for each macroblock, and using the second encoding method. And an encoding step for generating encoded image data.

A program according to the present invention is a computer-readable program for executing an image coding method conversion method, wherein image data encoded by the first coding method is a second code different from the first coding method. An image encoding method conversion method for converting into image data encoded by an encoding method, wherein the image data encoded by the first encoding method is decoded to generate decoded image data and a macroblock Decoding process for detecting each coding coefficient and converting the coding coefficient of the first coding method for each macroblock detected in the decoding process to the coding coefficient of the second coding method And encoding the decoded image data using the encoding coefficient of the second encoding method obtained by converting the decoded image data in the encoding coefficient conversion process for each macroblock. Coding It is characterized by comprising: a coding step of generating the image data encoded by the expression, a.

Brief Description of Drawings

FIG. 1 is a block diagram showing the configuration of an image coding method conversion apparatus according to the present invention. FIG. 2 is a diagram showing a quantization coefficient conversion table used in the apparatus of FIG. FIG. 3 is a diagram showing an example of image data used for creating a quantization coefficient conversion table. FIG. 4 is a block diagram showing the configuration of the encoding apparatus in the apparatus of FIG. FIG. 5 is a flowchart showing the encoding method conversion operation of the apparatus of FIG.

FIG. 6 is a flowchart showing the encoding process by the encoding device.

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of an image coding method conversion apparatus according to the present invention. The image encoding method conversion apparatus shown in FIG. 1 converts image data encoded by the MPEG 2 method as the first encoding method into H.264-based encoded image data of the second encoding method. The device includes an input buffer 1, a decoding device 2, a frame memory 3, a rate controller 4, an encoding device 5, and a transmission buffer 6.

The input buffer 1 is a memory for storing a stream of input image data encoded by the MPEG 2 method before decoding. The decoding device 2 decodes the image data stored in the input buffer 1 and supplies the decoded image data to the frame memory 3. At the time of decoding, a motion vector and a quantization coefficient are extracted for each macroblock and a picture code amount is detected. The motion vector information is supplied to the encoding device 5, and the quantization coefficient and picture code amount information is supplied to the rate controller 4. The frame memory 3 holds at least one frame of image data decoded by the decoding device 2.

The rate controller 4 controls the quantization coefficient at the time of quantization in the encoding device 5 based on the quantization coefficient and the information of the picture code amount supplied from the decoding device 2. The encoding device 5 converts the decoded image data held in the frame memory 3 into H.264-type encoded image data. A quantization coefficient conversion table is used for conversion. The quantization coefficient conversion table is for MPEG 2 quantization coefficient G. The quantization coefficient Q of H.264 system corresponds, and when expressed as a function, there is a relationship of Q = f (G). An example of the quantization coefficient conversion table is shown in FIG. The range of quantization coefficients in MPEG 2 (first predetermined range) is 1 to 32, and the quantization scale determined for each quantization coefficient differs between linear scale mode and nonlinear scale mode. The S / N ratio of decoded coefficients and decoded images is not related. On the other hand, the quantization coefficient range (second predetermined range) of H.264 is 0 to 51, and the quantization scale is set so that the quantization coefficient and the SZN ratio of the decoded image are proportional. It has been established. In order to create a quantization coefficient conversion table for associating different quantization coefficients for each method in this way, specific image data as shown in Fig. 3 is encoded using the MP EG 2 method while changing the quantization coefficient. The difference d between the reconstructed image A obtained by decoding the decoded image A and the reconstructed image B obtained by encoding the same image data in the same way using the H.264 method and decoding it is as follows. Calculated by the formula. Incidentally, a _xy elements of the reproduced image A obtained by decrypt, b _xy are elements of the reproduced image B obtained by decoding.

A quantization coefficient conversion table is created by associating quantization coefficients with a small difference d.

The transmission buffer 6 holds the encoded image data output from the encoding device 5 and outputs it to a predetermined transmission path.

As shown in Fig. 4, the encoding device 5 includes an arithmetic unit 11, a DCT converter 12, a quantizer 13, a variable length encoder 14, a buffer 15, an inverse quantizer 16, and an inverse DCT transform vessel 1 7, Computation unit 1 8, Frame memory 1 9, Motion compensator 2 0 and Motion vector holder 2 1

The arithmetic unit 11 subtracts the predicted image data motion-compensated by the motion compensator 20 for each macroblock from the input image data, and outputs the difference data to the DCT converter 12. Macroblock is each block obtained by dividing image data into 16 pixels x 16 pixels. Note that the processing block for converting the input image data into macroblocks is not shown in FIG. The 00 converter 1 2 performs two-dimensional discrete cosine transform on the output data of the arithmetic unit 1 1. The quantizer 13 quantizes the image data after DCT conversion and outputs the quantized image data to the variable length encoder 14 and the inverse quantizer 16. The variable length encoder 1 4 performs variable length encoding on the quantized data supplied from the quantizer 1 3 and the motion vector supplied from the decoding device 2 and outputs the result to the no-offer 15. . The buffer 15 holds and outputs the variable length data supplied from the variable length encoder 14. The inverse quantizer 16 inversely quantizes the quantized data input from the quantizer 13 and outputs it to the inverse DCT converter 17. The inverse D C T converter 17 performs inverse D C T conversion on the dequantized data and supplies it to the computing unit 18. The arithmetic unit 18 adds the predicted image data motion-compensated by the motion compensator 20 and the error data input from the inverse DCT converter 17 to convert to the original image data. Supply the image data to the frame memory 19 and store it.

The motion compensator 20 generates predicted image data by performing motion compensation on the image data read from the frame memory 19 corresponding to the motion vector. The motion vector holder 2 1 holds and outputs the motion vector information supplied from the decoding device 2. The operation of each unit 1 1 to 2 1 of the encoding device 5 is a controller (not shown). Controlled by.

Next, the image coding method conversion operation by the image coding method conversion having such a configuration will be described with reference to the flowcharts of FIGS. FIG. 5 shows an outline of the system conversion operation by the image encoding system conversion apparatus, and FIG. 6 shows an outline of the encoding process operation by the encoding apparatus 5. In the charts of Figs. 5 and 6, thick arrow lines indicate the flow of data, and thin arrow lines indicate the flow of processing and requests. Here, it is assumed that the encoded image data of the MPEGG2 system of frame k is supplied to the buffer 1.

The inputted MPE G 2 encoded image data of frame k is supplied from the buffer 1 to the decoding device 2 for each macroblock. The decryption device 2 performs decryption processing on the MPEG 2 system and outputs decoded image data (step S .1). The output decoded image data is held in the frame memory 3 (step S 2). In the decoding process, a motion vector and a quantization coefficient are extracted for each macroblock and a picture code amount is detected. The motion vector information is supplied to the encoder 5 and held in the motion vector holder 21. Information on the quantization coefficient and the picture code amount is supplied to the rate controller 4 and held in an internal memory (not shown) (step S 3).

The encoding device 5 executes an encoding process for converting the decoded image data output from the decoding device 2 into H.264-type encoded image data (step S 4).

In the encoding process, specifically, as shown in FIG. 6, the image indicated by the decoded image data of the supplied frame k is a macro block composed of 16 pixels × 16 pixels. Is divided into n blocks (step SI 01), and the counter value i is set to 0 (step S 102). Then, the quantization coefficient for macroblock i is set (step S 1 03). In step S 1 03, the quantization coefficient of macroblock i is required for rate controller 4.

In response to the request for the quantization coefficient, the rate controller 4 reads the quantization coefficient of the macroblock i from the internal memory as G (step S5), and calculates the quantization coefficient Q corresponding to the quantization coefficient G as the quantization coefficient. Quantization coefficient conversion is performed by searching and reading from the conversion table (step S6). The quantization coefficient Q is designated as the quantization coefficient of the macroblock i (step S7).

When the quantization coefficient Q for the macroblock i is set, it is determined whether or not the frame k is an intra-coded picture (step S 104). Information about whether or not the frame is an intra-coded picture is supplied from the decoding apparatus 2 together with the motion vector information. If the picture is not an intra-coded picture, that is, if it is an inter-coded picture, the motion vector of macroblock i is read from the motion vector holder 21 (step S 105), and the frame memory 1 The image data stored in 9 is read out, motion compensation is performed, and predicted image data is output (step S 106). The MP EG 2 motion vector predicts with 1Z2 pixel accuracy, and the H.264 motion vector has 1Z4 pixel accuracy, so the MP EG 2 motion vector must be doubled. The scaling is performed by Predicted image data is supplied to computing units 1 1 and 1 8, and computing unit 1 1 supplies differential data (prediction error) between image data of macroblock i and predicted image data to DCT converter 1 2. DCT converter 1 2 DCT converts the difference data The data is supplied to the quantizer 1 3 (step S 107), and the quantizer 1 3 quantizes the supplied data in accordance with the set quantization coefficient, and the quantized data is converted into the variable length encoder 14 and This is supplied to the inverse quantizer 16 (step S 1 08).

If it is determined in step S 104 that the picture is an intra-encoded picture, the image data of the macroblock i is transferred to the DCT converter 1 via the arithmetic unit 1 without executing steps S 1 05 and S 106. Supplied to 2. Therefore, in step S107, not the difference data but the image data of the macro block i is DCT transformed, and in step S108, the image data after the DCT transformation is quantized. The inverse quantizer 16 dequantizes the quantized data supplied from the quantizer 13 according to the set quantization coefficient (step S 109), and the inverse DC T converter 17 is inversely quantized. The obtained data is further subjected to inverse DCT conversion and supplied to the arithmetic unit 18 (step S 1 1 0). The arithmetic imager 18 is supplied with the predicted image data from the motion compensator 20 together with the inverse DCT conversion result data. The arithmetic unit 18 adds the data and decodes the image data for each macro block. This decoded image data is stored in the frame memory 19 in units of frames in step S 114 described later, and is used as a reference image for encoding the image data of the next frame.

The variable length encoder 14 performs variable length encoding on the quantized data supplied from the quantizer 1 3 and the motion vector supplied from the motion vector holder 2 1, Output (Step S 1 1 1). The buffer 1'5 holds the variable length data supplied from the variable length encoder 14 as H.264 encoded image data. Each time the above steps S103 to S1111 are processed, the counter value i is incremented by 1 (step S112). And the counter value i is 1 frame It is determined whether the number of blocks is smaller than n (step S 1 1 3). If i <n, the process returns to step S 1 03 and the encoding process for the next macroblock i is performed. If i≥n, the encoding process for frame k is completed, and the decoded image data for frame k is stored in frame memory 19 (step S 114). After the encoding process is completed, the decoded image data of frame k held in the buffer 15 is output via the transmission buffer 6 (step S8). Thereafter, returning to the above step S 1, the above-described image coding method conversion operation is performed on the MP EG 2 coded image data of the next frame k (= k + 1).

As described above, in the image encoding method conversion operation by the image encoding method conversion apparatus, the H.264 method quantization coefficient Q corresponding to the MPEG-2 method quantization coefficient G obtained by the decoding process is used. Is read and set for each macroblock from the quantization coefficient conversion table, and quantization is performed in the re-encoding process according to the quantization coefficient Q. Therefore, the quantization coefficient setting in the re-encoding process can be simplified. In addition, since the quantization coefficient Q of the H.264 system corresponding to the quantization coefficient G of the MP EG 2 system is set using the quantization coefficient conversion table, image quality degradation due to re-encoding of the H.264 system is reduced. Can be prevented.

In the above-described embodiment, the quantization coefficient is shown as the encoding coefficient used for encoding by the encoding means. However, the present invention is not limited to this, and other coefficients used for encoding may be used. . .

Furthermore, it can be provided as a computer program that executes the same operation as the encoding method conversion operation by the above-described image encoding method conversion apparatus.

As described above, according to the present invention, image data encoded by the first encoding method is recovered. Decoding means for generating decoded image data and detecting coding coefficients for each macroblock, and coding coefficients for the first coding method for each macroblock detected by the decoding means Coding coefficient conversion means for converting the image data into coding coefficients of the second coding method, and coding coefficients of the second coding method obtained by converting the decoded image data for each macroblock by the coding coefficient conversion means. And encoding means for generating image data encoded by the second encoding method by using the encoding method, so that deterioration of image quality due to re-encoding by a different encoding method can be prevented. .

Claims

The scope of the claims

1. An image encoding method conversion device for converting image data encoded by a first encoding method into image data encoded by a second encoding method different from the first encoding method,

Decoding means for decoding image data encoded by the first encoding method to generate decoded image data and detecting an encoding coefficient for each macroblock; and the decoding means Coding coefficient conversion means for converting the coding coefficient of the first coding method for each detected macroblock into the coding coefficient of the second coding method;

The decoded image data was encoded using the encoding coefficient of the second encoding method converted by the encoding coefficient conversion means for each macroblock and encoded by the second encoding method. An image encoding method conversion apparatus characterized by comprising: encoding means for generating image data.

2. The decoding means detects a quantized coefficient as an encoded coefficient for each macroblock,

The coding coefficient conversion means converts the quantization coefficient of the first coding scheme detected by the decoding means to the quantization coefficient of the second coding scheme for each macro block,

The encoding means includes quantization means for quantizing the decoded image data in accordance with a quantization coefficient of the second encoding method converted by the encoding coefficient conversion means for each macroblock. 2. The image coding method conversion device according to claim 1, wherein

3. The coding coefficient conversion means includes: a first predetermined range of quantization coefficients of the first coding scheme; and a second predetermined range of quantization coefficients different from the first predetermined range of the second coding scheme; Means for storing a quantization coefficient conversion table indicating the correspondence between

Reading means for reading out the quantized coefficient of the second encoding method corresponding to the quantized coefficient of the first encoding method for each macroblock detected by the decoding means from the quantized coefficient conversion table; The image coding method conversion apparatus according to claim 2, further comprising:

4. The image coding method conversion apparatus according to claim 1, wherein the first coding method is M P E G 2 and the second coding method is H.

5. The decoding means detects a motion vector for each macroblock when decoding the image data encoded by the first encoding method,

The encoding means has motion compensation means for performing motion compensation using the detected motion vector for each macroblock when the decoded image data is encoded by the second encoding method. The image coding system conversion device according to claim 1 or 2.

6. An image encoding method conversion method for converting image data encoded by the first encoding method into image data encoded by a second encoding method different from the first encoding method,

Decoding process for decoding image data encoded by the first encoding method to generate decoded image data and detecting an encoding coefficient for each macroblock, and detecting in the decoding process A coding coefficient conversion step for converting the coding coefficient of the first coding method into the coding coefficient of the second coding method for each McPlock block; An image obtained by encoding the decoded image data using the encoding coefficient of the second encoding method converted in the encoding coefficient conversion process for each macro block and encoding the encoded image data by the second encoding method An image encoding method conversion method comprising: an encoding step for generating data;

7. A computer-readable program for executing an image coding method conversion method,

An image encoding method conversion method for converting image data encoded by a first encoding method into image data encoded by a second encoding method different from the first encoding method,

Decoding image data encoded by the first encoding method to generate decoded image data, and detecting a coding coefficient for each macroblock; and detecting in the decoding step An encoding coefficient conversion step of converting the encoding coefficient of the first encoding method for each macroblock into the encoding coefficient of the second encoding method;

An image obtained by encoding the decoded image data using the encoding coefficient of the second encoding method converted in the encoding coefficient conversion process for each macro block and encoding the encoded image data by the second encoding method An encoding process for generating data, and a program comprising: