JP2009060320A - Image processing device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing device capable of improving the ease of access to a specific picture element contained in an image data after encoding and offering an image suppressed in picture deterioration. <P>SOLUTION: An image processing device 100 comprises a wavelet transformation portion 102 for performing wavelet transformation to an image block captured by an image capturing portion 101, a first entropy encoding portion 103 for encoding a high-pass factor after wavelet transformation, a second entropy encoding portion 105 for encoding a low-pass factor after wavelet transformation, a code amount determining portion 106 for determining the generated code amount of entropy code data, a quantization processing portion 107 for performing quantization and encoding processing to the image block, and a code output portion 108 for outputting code data. The code output portion 108 selects either the entropy code data or the quantized data according to whether the generated code amount is within a target code amount or not, and outputs code data of the target code amount to the image block. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and method, and more particularly to an image processing apparatus that compresses image data and an image processing method using the apparatus.

  Conventionally, various methods are used for compression (encoding) of image data. Representative examples include MPEG, JPEG, JPEG2000, etc., which are international standards. In MPEG or JPEG, when image data compression processing is performed, the image data is divided into blocks of a certain size. This block generally has a size of 16 × 16 pixels or 8 × 8 pixels. It is known that block noise occurs due to the effect of compression distortion when such processing in units of blocks is performed. On the other hand, in JPEG2000, in order to prevent the occurrence of block noise, a method that can process the entire image as one block is adopted.

  By performing these image compressions, the data size can be reduced to a small size such as a fraction of 1 to several tenths compared to the original image data. However, in order to perform such compression processing and, on the contrary, decompression processing for returning from compressed data to image data, a memory for holding image data and a buffer memory for holding data being processed are provided. There is a tendency to need more. As a method for reducing the buffer memory in the middle of such processing, for example, a method as in Patent Document 1 has been proposed.

In this patent document 1, in particular, when processing is performed by dividing an image into blocks called tiles (including the case where the image is not divided) in JPEG2000 processing, the memory required for the processing step is managed to manage the memory. Techniques for improving utilization efficiency are described.
JP 2006-287485 A

  In the case of image data compressed by the above-described method, for example, when accessing a specific pixel included in this image data in association with image processing such as IP conversion and image quality correction, a code of a block including at least the pixel It is necessary to decompress the data. When the entire image is encoded as one block as in JPEG2000, it is necessary to expand the entire image.

  In compression processing such as MPEG / JPEG / JPEG2000, entropy encoding is used, and the code amount of encoded data corresponding to each block is almost constant. For this reason, in order to extract a certain pixel in the image data, the block including the pixel must be expanded. However, in order to extract the compressed data of the block including the pixel from the compressed image data, Processing such as searching for compressed data of a target block while sequentially reading header information from the head of the compressed data is required, and there is a problem that accessibility is poor.

  The present invention has been made in view of the above-described circumstances, and can improve the accessibility to specific pixels included in encoded image data and can reduce image quality degradation. It is an object of the present invention to provide an image processing method using the apparatus.

  In order to solve the above-described problem, a first technical means of the present invention is an image processing apparatus that compresses image data, and that extracts an image block of a predetermined size from image data to be compressed. Means, wavelet transform processing means for performing wavelet transform processing on the clipped image block, first encoding means for entropy encoding high frequency coefficients output from the wavelet transform processing means, and the wavelet Differential encoding means for differentially encoding the low frequency coefficient output from the transform processing means; second encoding means for entropy encoding the output of the differential encoding means; the first encoding means; and A code amount determining means for determining a generated code amount from an output of the second encoding means; and an image block cut out by the image cutting out means. Quantization means for performing quantization and encoding processing on the image and output means for outputting code data based on the determination result by the code amount determination means, wherein the output means is determined by the code amount determination means As a result, depending on whether or not the generated code amount falls within a predetermined target code amount, either the entropy code data or the quantized data is selected, and the target code amount is selected for the image block. The code data is output.

  According to a second technical means, in the first technical means, the quantization means performs quantization and encoding so that a code amount for each image block matches the target code amount, and the output means includes: When entropy code data is output, an output that matches the target code amount is performed.

  The third technical means is characterized in that, in the first or second technical means, the image block cut out by the image cut-out means is composed of data of a predetermined number N pixels which are horizontally continuous. is there.

  According to a fourth technical means, in the first or second technical means, the first encoding means and / or the second encoding means performs a Huffman encoding process. .

  A fifth technical means is the first or second technical means, wherein the first encoding means and / or the second encoding means performs arithmetic encoding processing. .

  A sixth technical means is any one of the first to fifth technical means, wherein the output means selects the entropy code data when the generated code amount is within a predetermined target code amount, and the generation When the code amount exceeds a predetermined target code amount, the quantized data is selected.

  A seventh technical means is an image processing method by an image processing apparatus that compresses image data, an image cutout step of cutting out an image block of a predetermined size from image data to be compressed, and the cutout A wavelet transform processing step for performing wavelet transform processing on the image block, a first encoding step for entropy encoding the high frequency coefficient output in the wavelet transform processing step, and output in the wavelet transform processing step A differential encoding step for differentially encoding the low frequency coefficients, a second encoding step for entropy encoding the output in the differential encoding step, the first encoding step, and the second encoding A code amount determination step of determining a generated code amount from an output in the step; and the image A quantization step for performing quantization and encoding processing on the image block cut out in the extraction step; and an output step for outputting code data based on a determination result in the code amount determination step, the output The step selects code data of either entropy code data or quantized data according to whether or not the generated code amount falls within a predetermined target code amount as a result of the determination in the code amount determination step, The code data of the target code amount is output for the image block.

  According to an eighth technical means, in the seventh technical means, the quantization step performs quantization and encoding so that a code amount for each image block matches the target code amount, and the output step comprises: When entropy code data is output, an output that matches the target code amount is performed.

  According to a ninth technical means, in the seventh or eighth technical means, the image block cut out in the image cut-out step is composed of data of a fixed number N pixels which are horizontally continuous. It is.

  A tenth technical means is the seventh or eighth technical means, wherein the first encoding step and / or the second encoding step performs Huffman encoding processing. .

  The eleventh technical means is characterized in that, in the seventh or eighth technical means, the first encoding step and / or the second encoding step performs arithmetic encoding processing. .

  In a twelfth technical means according to any one of the seventh to eleventh technical means, in the output step, when the generated code amount is within a predetermined target code amount, the entropy code data is selected and the generation is performed. When the code amount exceeds a predetermined target code amount, the quantized data is selected.

  According to the present invention, since a certain amount of code data is output for each image block, when it is desired to access a pixel at a specific position in the image data, the position of the code data including the pixel can be easily specified. Thus, it is possible to access the pixel at high speed while suppressing image quality deterioration.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of an image processing apparatus and an image processing method using the apparatus according to the invention will be described with reference to the accompanying drawings.

  FIG. 1 is a block diagram illustrating a configuration example of an image processing apparatus according to an embodiment of the present invention. In FIG. 1, reference numeral 100 denotes the image processing apparatus. The image processing apparatus 100 includes an image cutout unit 101 that corresponds to an image cutout unit that cuts out an image block having a predetermined size from image data to be compressed, and a wavelet transform on the image block cut out by the image cutout unit 101. And a wavelet transform unit 102 corresponding to wavelet transform processing means for performing the above.

  Furthermore, the image processing apparatus 100 includes a first entropy encoding unit 103 corresponding to a first encoding unit that performs entropy encoding on the high-frequency coefficient after wavelet transform, and a low-frequency coefficient after wavelet transform. A pre-processing unit 104 corresponding to a differential encoding unit that performs differential encoding, and a second encoding unit that performs entropy encoding on code data subjected to the differential encoding process An entropy encoding unit 105, and a code amount determination unit 106 corresponding to a code amount determination unit that determines a generated code amount of code data output from the first entropy encoding unit 103 and the second entropy encoding unit 105; Based on the determination result of the quantization processing unit 107 corresponding to the quantization means for performing quantization and encoding processing on the image block, and the code amount determination unit 106 And a code output unit 108 corresponding to the output means for outputting the data.

  The main characteristic part of the present invention is that the code output unit 108 generates the code of the code data output from the first entropy encoding unit 103 and the second entropy encoding unit 105 as a result of the determination by the code amount determination unit 106. Depending on whether or not the amount falls within a predetermined target code amount, either the entropy code data or the quantized data is selected, and the code data of the target code amount is output to the image block. is there.

  By adopting such a configuration, the image data to be compressed is cut out in units of a predetermined small image block and encoded, and the code amount by encoding is controlled to a predetermined value, A certain amount of encoded data is also output for each pixel block. For this reason, for example, when it is necessary to perform image processing such as IP conversion and image quality correction, and when it is desired to access a pixel at a specific position in the image data, the data amount of each pixel block is constant. Since the position of data can be easily specified, high-speed access is possible.

  Further, when the target code amount can be achieved by the entropy encoding process, lossless compression is performed. However, when only the lossless compression is performed, the target code amount may be exceeded. On the other hand, in the case of quantized data, since it is irreversible compression, the amount of code can be reduced. However, since it involves image quality deterioration, it is desirable to use entropy code data that is lossless compression as much as possible. Therefore, as long as the code amount of entropy code data does not exceed the target code amount, lossless compression entropy code data is output, and only when the code amount exceeds the target code amount, lossy compression quantized data is output. I have to. This makes it possible to prevent image quality deterioration while suppressing the code amount.

  FIG. 2 is a diagram for explaining an example of an image block cut out by the image cutout unit 101. In FIG. 2, 201 denotes an entire image to be compressed, and 202 denotes an example of an image block to be cut out. As described above, the image data 201 to be compressed is cut out into the image block 202 by the image cutout unit 101. In this example, data of N pixels (a fixed number) that are horizontally continuous are cut out as one block. As N, an appropriate value can be selected in consideration of the size of the original image data, accessibility to the data after compression, such as 32, 64, 80, 100, and 128. In consideration of the next wavelet transform process, N is preferably an even number.

Next, the wavelet transform unit 102 performs wavelet transform on the image block cut out by the image cutout unit 101. As a result, the image data is divided into a high frequency coefficient and a low frequency coefficient.
FIG. 3 is a diagram schematically showing a state in which an image block is divided into a high frequency coefficient and a low frequency coefficient by the wavelet transform unit 102. In FIG. 3, 301 is a pixel (image) block, 302 is a low frequency coefficient, and 303 Indicates a high frequency coefficient. In this way, the pixel block 301 is separated into a low frequency coefficient 302 and a high frequency coefficient 303 by wavelet transformation.

  Next, the high frequency coefficient 303 is encoded by the first entropy encoding unit 103. As the encoding method, for example, a Huffman code, an arithmetic code, or the like can be used. The first entropy encoding unit 103 sends the encoding result here to the code output unit 108 and notifies the code amount determination unit 106 of the generated code amount.

  The low frequency coefficient 302 is preprocessed by the preprocessing unit 104. Here, differential encoding (DPCM: Differential Pulse Code Modulation) is performed. By performing this differential encoding, it is expected that the data value is unevenly distributed in the vicinity of 0 due to the similarity of adjacent pixels. Therefore, an improvement in efficiency in the next entropy encoding can be expected. The output (code data) of the preprocessing unit 104 is encoded by the second entropy encoding unit 105. Again, for example, a Huffman code or an arithmetic code can be used as the encoding method. The second entropy encoding unit 105 sends the encoding result here to the code output unit 108 as described above, and notifies the code amount determination unit 106 of the generated code amount.

  A target code amount is given to the quantization processing unit 107 in advance. In parallel with the above-described wavelet transform to entropy encoding processing, the quantization processing unit 107 performs quantization and encoding processing on the image block cut out by the image cutout unit 101 so that the target code amount is obtained. Do. As the quantization process, for example, a process of truncating the lower bits of the pixel data so as to become the target code amount is performed, and as the encoding process, information indicating how many bits of the lower bits are truncated is added.

Also, the code amount determination unit 106 is given a target code amount in advance as in the quantization processing unit 107. The code amount determination unit 106, when the sum of the generated code amounts output from the first entropy encoding unit 103 and the second entropy encoding unit 105 is within the target code amount, entropy code data Is output to the code output unit 108. For the amount that the generated code amount does not reach the target code amount, for example, the code output unit 108 is instructed to output 0 data or the like until the target code amount is reached, and the output entropy code data becomes the target code amount. Adjust as follows. Thereby, the entropy code data can be output by the target code amount.
On the other hand, when the total generated code amount exceeds the target code amount, the code output unit 108 is instructed to output code data obtained as a result of processing by the quantization processing unit 107. Since the quantization processing unit 107 is adjusted so that the output quantized data has the target code amount, the quantized data can be output by the target code amount.

  Even if either entropy code data or quantized data is selected by the above processing, the code output unit 108 always outputs code data of a predetermined target code amount for each image block. Will be.

FIG. 4 is a flowchart for explaining an example of an image processing method by the image processing apparatus 100 shown in FIG. First, the image processing apparatus 100 cuts out an image block having a predetermined size from image data to be compressed (step S1), and performs wavelet transform processing on the cut out image block (step S2).
Then, the image processing apparatus 100 performs a first entropy encoding process for entropy encoding the high frequency coefficient after the wavelet transform (step S3), and differentially encodes the low frequency coefficient after the wavelet transform as a preprocess (step S3). S4) Further, a second entropy encoding process for entropy encoding is performed on the differentially encoded data (step S5).

  Next, in the image processing apparatus 100, the code amount determination unit 106 determines the generated code amount from the outputs (code data) of the first entropy encoding unit 103 and the second entropy encoding unit 105 (step S6). . The code amount determination unit 106 outputs the code data from the first entropy encoding unit 103 and the second entropy encoding unit 105 when the generated code amount is within the target code amount (in the case of YES). The sign output unit 108 is instructed (step S8). On the other hand, when the generated code amount exceeds the target code amount in step S6 (in the case of NO), the code output unit 108 is instructed to output the code data quantized by the quantization processing unit 107 (steps S7, 8). ). Even if either entropy code data or quantized data is selected, the code output unit 108 always outputs code data corresponding to a target code amount given in advance.

As described above, according to the present invention, image data to be compressed is cut out in units of a predetermined small image block and encoded, and the code amount by encoding is controlled to a predetermined value. Thus, a certain amount of encoded data is output for each pixel block. For this reason, when it is desired to access a pixel at a specific position in the image data, the position of the compressed data including the pixel can be easily specified, so that high-speed access is possible.
In addition, as long as the amount of code generated by entropy coding does not exceed the target code amount, entropy code data that is lossless compression is output, so high-speed access to a pixel at a specific position in the image data while suppressing image quality degradation Is possible.
Further, since it is only necessary to perform decompression processing on the compressed image data in units of small image blocks, a processing amount for obtaining necessary pixels can be reduced.

It is a block diagram which shows the structural example of the image processing apparatus which concerns on one Embodiment of this invention. It is a figure for demonstrating an example of the image block cut out by the image cutout part. It is the figure which showed typically the state which divided the image block into the high frequency coefficient and the low frequency coefficient by the wavelet transformation part. It is a flowchart for demonstrating an example of the image processing method by the image processing apparatus shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Image processing apparatus, 101 ... Image clipping part, 102 ... Wavelet transformation part, 103 ... 1st entropy encoding part, 104 ... Pre-processing (DPCM) part, 105 ... 2nd entropy encoding part, 106 ... code | symbol Amount determination unit 107 ... quantization processing unit 108 108 code output unit 201 image data 202 202 image block 301 pixel block 302 low frequency coefficient 303 high frequency coefficient

Claims (12)

An image processing apparatus that compresses image data,
Image cutout means for cutting out an image block of a predetermined size from image data to be compressed, wavelet transform processing means for performing wavelet transform processing on the cutout image block, and output from the wavelet transform processing means First encoding means for entropy-encoding the high-frequency coefficients, differential encoding means for differentially encoding the low-frequency coefficients output from the wavelet transform processing means, and an output of the differential encoding means for entropy coding Second encoding means to be converted, code amount determination means for determining a generated code amount from outputs of the first encoding means and the second encoding means, and an image block cut out by the image cut-out means Quantization means for performing quantization and encoding processing on the code, and a code based on the determination result by the code amount determination means And output means for outputting the over data,
The output means selects code data of either entropy code data or quantized data depending on whether or not the generated code quantity is within a predetermined target code quantity as a result of the determination by the code quantity determination means And outputting the code data of the target code amount for the image block.   The quantization means performs quantization and encoding so that the code amount for each image block matches the target code amount, and the output means matches the target code amount when outputting entropy code data. The image processing apparatus according to claim 1, wherein output is performed.   The image processing apparatus according to claim 1 or 2, wherein the image block cut out by the image cut-out means is composed of data of a fixed number of N pixels that are horizontally continuous.   The image processing apparatus according to claim 1, wherein the first encoding unit and / or the second encoding unit performs a Huffman encoding process.   The image processing apparatus according to claim 1, wherein the first encoding unit and / or the second encoding unit performs an arithmetic encoding process.   The output means selects entropy code data when the generated code amount is within a predetermined target code amount, and selects quantized data when the generated code amount exceeds a predetermined target code amount. The image processing apparatus according to any one of claims 1 to 5. An image processing method by an image processing apparatus that compresses image data,
An image cutout step for cutting out an image block of a predetermined size from image data to be compressed, a wavelet transform processing step for performing wavelet transform processing on the cutout image block, and output at the wavelet transform processing step A first encoding step that entropy-encodes the high-frequency coefficients, a differential encoding step that differentially encodes the low-frequency coefficients output in the wavelet transform processing step, and an output in the differential encoding step It is cut out in a second encoding step for entropy encoding, a code amount determination step for determining a generated code amount from outputs in the first encoding step and the second encoding step, and the image cutout step. The amount to quantize and encode the image block It comprising of a step, and an output step of outputting the code data based on the determination result of the code amount determining step,
The output step selects either the entropy code data or the quantized data depending on whether or not the generated code amount is within a predetermined target code amount as a result of the determination in the code amount determination step. And outputting code data of the target code amount for the image block.   The quantization step performs quantization and encoding so that the code amount for each image block matches the target code amount, and the output step matches the target code amount when entropy code data is output. The image processing method according to claim 7, wherein output is performed.   9. The image processing method according to claim 7, wherein the image block cut out in the image cut-out step is composed of data of a fixed number of N pixels that are horizontally continuous.   The image processing method according to claim 7 or 8, wherein the first encoding step and / or the second encoding step performs a Huffman encoding process.   The image processing method according to claim 7 or 8, wherein the first encoding step and / or the second encoding step performs an arithmetic encoding process.   The output step selects the entropy code data when the generated code amount is within a predetermined target code amount, and selects the quantized data when the generated code amount exceeds a predetermined target code amount. The image processing method according to claim 7, wherein the image processing method is any one of claims 7 to 11.

Images