JP2006332756A - Image decoder - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To decrease the processing amount of an image decoder for receiving moving picture data including pixel data and quantization table data by each frame. <P>SOLUTION: A memory 12 stores the quantization table data of a frame received at first as default table data, and the image decoder attains decode processing of a plurality of frames by using the default table data. A table comparison section 24 compares only low frequency components of the quantization table data acquired at a current frame with those of the default table data and updates the default table data only when the comparison indicates a result of dissidence in excess of a prescribed limit. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image decoding apparatus that decodes encoded image information.

  JPEG (Joint Photographic Coding Experts Group) technology, which is one of the still image compression methods, obtains data (DCT coefficients) converted into frequency components by DCT (Discrete Cosine Transform) and is specified in the JPEG stream. Data encoding is realized by rounding the DCT coefficient using the quantization table. A moving image coding method in which a JPEG stream is continuously recorded for each frame is called motion JPEG (Motion-JPEG).

According to a conventional technique, JPEG expansion processing is performed by supplying quantization table data in which the quantization coefficient of a high-frequency component is zero. As a result, a smoothed reproduction image from which high-frequency components have been cut is obtained, so that it is not necessary to perform a smoothing process separately, and the processing time is greatly shortened. This is based on the fact that the high frequency components of the quantization table do not significantly affect the image (see Patent Document 1).
JP-A-10-341369

  It is rare that the quantization table in the motion JPEG stream is different for each frame, and conventionally, a quantization table is obtained for each frame and stored, resulting in a waste of processing amount.

  An object of the present invention is to reduce the processing amount of an image decoding apparatus that receives moving image data having pixel data and quantization table data for each frame.

  In order to achieve the above object, the present invention stores quantization table data of a frame first input to an image decoding apparatus as default table data, and performs decoding processing of a plurality of frames using the default table data. It is to achieve.

  The default table data may be updated when a mismatch exceeding a predetermined limit is detected as a result of comparing only the low frequency components between the quantization table data acquired in the current frame and the default table data.

  According to the present invention, the processing amount of the image decoding apparatus can be reduced by using the default table data.

  Hereinafter, embodiments of an image decoding apparatus according to the present invention will be described with reference to the drawings.

  FIG. 1 shows a configuration example of an image decoding apparatus according to the present invention. The image decoding apparatus in FIG. 1 is a motion JPEG stream decoding apparatus, and includes an image information separation unit 10, an initial quantization table separation unit 11, a default table data memory 12, a pixel data decoding unit 13, and a default table. The data update unit 20 is configured. The default table data update unit 20 includes a quantization table frequency component separation number determination unit 21, a quantization table frequency component separation unit 22, a default table frequency component separation unit 23, and a quantization table comparison unit 24. Yes.

  The input moving image data has pixel data and quantization table data for each frame. The pixel data is encoded data using DCT and Huffman encoding. Here, it is assumed that one macroblock is composed of 8 × 8 (= 64) pixels. The quantization table data is data used for quantization of 8 × 8 DCT coefficients, and includes 8 × 8 quantization coefficients.

  The image information separation unit 10 separates pixel data and quantization table data from the input moving image data. Pixel data is supplied to the pixel data decoding unit 13, and quantization table data is supplied to the initial quantization table separation unit 11. The initial quantization table separation unit 11 separates quantization table data included in a frame first input to the image decoding apparatus as default table data, and supplies the default table data to the default table data memory 12. The default table data memory 12 stores the default table data supplied from the initial quantization table separation unit 11. The pixel data decoding unit 13 decodes the pixel data supplied from the image information separation unit 10 based on the default table data stored in the default table data memory 12, and outputs the decoded data. Specifically, the pixel data decoding unit 13 performs Huffman decoding processing, inverse quantization processing, and inverse DCT processing.

  On the other hand, new quantization table data possessed by frames after the frame first input to the image decoding apparatus is supplied to the default table data update unit 20 via the initial quantization table separation unit 11. The default table data update unit 20 compares the new quantization table data supplied from the initial quantization table separation unit 11 and the default table data stored in the default table data memory 12 as a result of the comparison of only low frequency components, It has a function of updating the default table data memory 12 when a mismatch exceeding a predetermined limit is detected. Therefore, the quantization table frequency component separation unit 22 separates a predetermined number of low frequency components from the new quantization table data supplied from the initial quantization table separation unit 11. The default table frequency component separation unit 23 separates the same number of low frequency components from the default table data stored in the default table data memory 12. The quantization table frequency component separation number determination unit 21 determines the number of low frequency components to be separated by the quantization table frequency component separation unit 22 and the default table frequency component separation unit 23 for each frame. Here, the number of low frequency components is N, and the number of high frequency components is M. N and M are natural numbers of 2 or more, and are variable values whose values increase and decrease according to the frame size. As described above, when the quantization table data is composed of 64 (= 8 × 8) quantization coefficients, M = 64−N is established. For example, N = 10 and M = 54. The quantization table comparison unit 24 compares the low frequency component separated by the quantization table frequency component separation unit 22 with the low frequency component separated by the default table frequency component separation unit 23. As a result of the comparison in the quantization table comparison unit 24, when a mismatch exceeding a predetermined limit is detected in the ten low frequency components, the default table data stored in the default table data memory 12 is newly quantized. Updated to table data.

  As described above, according to the image decoding apparatus in FIG. 1, the quantization table data of the first input frame is stored as default table data, and a decoding process of a plurality of frames is achieved using the default table data. Therefore, the processing amount can be greatly reduced. Moreover, as a result of comparing only the low frequency components of the quantization table data acquired in the current frame and the default table data, the default table data is updated only when a mismatch exceeding a predetermined limit is detected. The processing amount is further reduced. The high-frequency component of the quantization table data is a part that does not affect the image so much, and the motion JPEG stream has a relatively high compression rate, so that there is no significant effect on the image even if the high-frequency component is not compared.

  FIG. 2 is a flowchart showing an example of processing of the quantization table frequency component separation number determination unit 21 in FIG. According to FIG. 2, the number N of low frequency components to be separated is limited to a predetermined low frequency component limit number L (for example, 8) or less. In the case of a color image, the number of low frequency components to be separated may be determined individually for each color component. For example, according to the YUV format, luminance table (Y) quantization table data and color difference signal (Cb, Cr) quantization table data are provided for each frame. In FIG. 2, Ny is the number of low frequency components in the luminance signal, and Nc is the number of low frequency components in the color difference signal.

  FIG. 3 is a flowchart showing an example of processing of the quantization table comparison unit 24 in FIG. According to FIG. 3, the quantization table comparison unit 24 calculates the difference between the low frequency component separated by the quantization table frequency component separation unit 22 and the low frequency component separated by the default table frequency component separation unit 23. Are all within the comparison threshold value K (for example, 10), the default table data memory 12 is controlled not to be updated.

  Note that the input moving image data of the image decoding apparatus according to the present invention is not limited to a motion JPEG stream. Further, the image decoding apparatus in FIG. 1 can be realized by either hardware or software.

  As described above, the image decoding apparatus according to the present invention can reduce the processing amount by using the default table data, and thus is useful as a motion JPEG stream decoding apparatus or the like.

It is a block diagram which shows the structural example of the image decoding apparatus which concerns on this invention. It is a flowchart which shows the example of a process of the quantization table frequency component isolation | separation number determination part in FIG. It is a flowchart which shows the example of a process of the quantization table comparison part in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image information separation part 11 Initial quantization table separation part 12 Default table data memory 13 Pixel data decoding part 20 Default table data update part 21 Quantization table frequency component separation number determination part 22 Quantization table frequency component separation part 23 Default table frequency Component separation unit 24 Quantization table comparison unit

Claims (7)

An image decoding apparatus that receives moving image data having pixel data and quantization table data for each frame,
An image information separation unit for separating pixel data and quantization table data from the moving image data;
An initial quantization table separation unit that separates quantization table data of a frame first input to the image decoding device as default table data;
A default table data memory for storing the default table data;
An image decoding apparatus comprising: a pixel data decoding unit that decodes the pixel data based on the default table data and outputs the decoded data as decoded data. The image decoding device according to claim 1,
When a mismatch exceeding a predetermined limit is detected as a result of comparing only the low frequency components of the new quantization table data and the default table data of the frames after the frame first input to the image decoding device The image decoding apparatus further comprises a default table data updating unit for updating the default table data memory. The image decoding device according to claim 2, wherein
The default table data update unit
A quantization table frequency component separation unit that separates a predetermined number of low frequency components from new quantization table data of frames subsequent to the frame first input to the image decoding device;
A default table frequency component separator for separating the same number of low frequency components from the default table data;
A quantization table comparison unit that compares the low frequency component separated by the quantization table frequency component separation unit and the low frequency component separated by the default table frequency component separation unit;
When a mismatch exceeding a predetermined limit is detected as a result of the comparison in the quantization table comparison unit, the default table data stored in the default table data memory is updated to the new quantization table data. An image decoding apparatus characterized by the above. The image decoding device according to claim 3, wherein
The default table data update unit includes a quantization table frequency component separation number determination unit that determines the number of low frequency components to be separated by the quantization table frequency component separation unit and the default table frequency component separation unit for each frame. An image decoding device further comprising: The image decoding device according to claim 4, wherein
The quantization table frequency component separation number determining unit has a function of limiting the number of low frequency components to be separated to a predetermined low frequency component limit number or less. The image decoding device according to claim 4, wherein
The quantization table frequency component separation number determination unit has a function of individually determining the number of low frequency components to be separated for each color component. The image decoding device according to claim 3, wherein
The quantization table comparison unit is configured such that all of the differences between the low frequency component separated by the quantization table frequency component separation unit and the low frequency component separated by the default table frequency component separation unit are within a comparison threshold value. If it is, the image decoding apparatus has a function of controlling so that the default table data memory is not updated.

Images