JP2000134639A - Encoding method for color still picture - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable efficient encoding by encoding a bottom frequency sub-band for each frame and performing zero tree encoding while coupling the quantizing values of respective frames to one concerning high frequency sub-bands. SOLUTION: The respective frames of luminance Y and color differences Cr and Cb in a color still picture are decomposed to plural sub-bands by wavelet transformation. In this case, when an image format is 4:2:0, the number of times of division for the color differences Cr and Cb is legs than that of luminance Y for one and when the image format is 4:2:2 or 4:4:4, the numbers of times of division to all the luminance Y and color differences Cr and Cb are made equal. Then, the bottom frequency sub-band is quantized respectively independently to the luminance Y and the color differences Cr and Cb and adaptive predictive encoding is performed. Further concerning the high frequency sub-band, the amplitude of a coefficient is quantized, and three quantized values of the luminance Y and color differences Cr and Cb are coupled into one and encoded by a zero tree. Thus, efficient encoding can be attained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color still image encoding method suitable for a still image communication and storage system, and more particularly to an efficient color image compression.

[0002]

2. Description of the Related Art Japanese Unexamined Patent Publication No. 10-164574 discloses "video compression". A still image is decomposed into a plurality of subbands by wavelet transform, the lowest subband is adaptively coded, and the high subband is a zero tree code. It describes a method to make it. When a color still image is encoded by this method, as shown in FIG.
Are independently encoded, and the bit streams are concatenated as follows, for example. | Y stream | C
r stream | Cb stream |.

[0003]

In encoding a color image, the conventional technique compresses information by removing the correlation in each frame of luminance Y, color difference Cr, and Cb.
The correlation between the three frames of Y, Cr, and Cb is not considered. Therefore, an object of the present invention is to provide a more efficient encoding method in consideration of the correlation between the luminance Y and the color difference Cr and Cb of a color still image.

[0004]

In order to solve the above-mentioned problems, the present invention decomposes each frame of luminance Y, color difference Cr and Cb of a color still image into a plurality of subbands, and It is intended to provide a color still image encoding method, wherein encoding is performed for each frame, and for a high frequency sub-band, the quantized value of each frame is combined into one and subjected to zero tree encoding. is there.

[0005]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing an embodiment of a color still image encoding method according to the present invention. First, each frame of luminance Y, color difference Cr, and Cb of a color still image is decomposed into a plurality of subbands by wavelet transform. However, when the image format is 4: 2: 0, the color difference C
Reduce the number of divisions of r and Cb by one less than that of luminance Y. If the image format is 4: 2: 2 or 4: 4: 4, luminance Y and color difference C
Make the number of all divisions of r and Cb equal. The lowest sub-band is independently quantized for luminance Y, chrominance Cr, and Cb, and is adaptively predictively coded.

For the high frequency sub-band, the amplitude of the coefficient is quantized, and the three quantized values of luminance Y, chrominance Cr, and Cb are combined into one by the method described below, and encoded by a zero tree. In this embodiment, two quantization values (Y and Cr, or Y and Cb)
And three quantized values (Y, Cr and Cb)
May be combined into one. First, the former will be described.

The two integers x and y are at most nx and ny, respectively.
Each bit value of x and y is xi (i = 0,..., Nx-1) and yj (j =
0, ..., ny-1). The following values are assigned to each bit.

[0008]

(Equation 1)

The corresponding bit values are arranged in ascending order of the values of px (i) and py (j) (when there are a plurality of equal values, priority is given to Y, Cr and Cb), and 1 of nx + ny bits is arranged.
Make two integers. The same applies to the case where three integers are combined into one.

Next, a description will be given of which of the quantized values of the wavelet coefficients are actually combined with which. 4: 2:
In the case of 0, the sizes of the Cr and Cb frames are half of the width and height of the Y frame, and the number of divisions of the Cr and Cb frames is one less than that of the Y frame. Therefore, as shown in FIG.
Since each component of the high frequency subbands other than the three highest frequency subbands of the Y frame can correspond to the components of the corresponding subbands of the Cr and Cb frames on a one-to-one basis, Y, C
The three corresponding components of r and Cb are combined to obtain one value. For the three highest subbands of the Y frame,
Since there is no corresponding Cr or Cb frame, the quantization value of the Y frame is used as it is.

In the case of 4: 2: 2, the size of the Cr and Cb frames is only half the width of the Y frame, and the number of divisions is the same for all frames. Therefore, as shown in FIG. 3, the high-frequency subbands of the Cr and Cb frames corresponding to the respective high-frequency subbands of the Y frame have a width of 1/2. Therefore, the component Qy (2i, j) having an even x coordinate of the Y frame is combined with the component Qcr (i, j) of the Cr frame, and the component Qy (2i + 1, j) having an odd x coordinate is a component of the Cb frame. Combine with the component Qcb (i, j) to create one value.

In the case of 4: 4: 4, since each component of all the high frequency sub-bands of the Y, Cr and Cb frames has a one-to-one correspondence, the corresponding three components are combined to form one value. .

As described above, according to the present embodiment, each frame of the luminance Y, color difference Cr, and Cb of the color still image is decomposed into a plurality of subbands, the lowest subband is encoded for each frame, and As for the sub-band, the quantization values of the respective frames are combined into one and zero-tree coding is performed, so that more efficient coding can be realized as compared with the conventional example.

In the wavelet coding of a color image, the coefficient at the corresponding position of the Y, Cr, and Cb frames is often an insignificant coefficient at the same time.
Since it can be represented by two codes, more efficient encoding can be performed as compared with the related art.

[0015]

As described above, the color still image encoding method of the present invention can realize more efficient encoding than the conventional technique.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an encoding method according to an embodiment.

FIG. 2 is a diagram showing correspondence between a Y frame and high-frequency subbands of Cr and Cb frames in the case of 4: 2: 0.

FIG. 3 is a diagram illustrating correspondence between a Y frame and a high-frequency subband of a Cr and Cb frame in the case of 4: 2: 2.

FIG. 4 is a block diagram for explaining a conventional encoding method.

Claims (1)

[Claims] 1. A frame of a luminance Y, a color difference Cr, and Cb of a color still image is decomposed into a plurality of subbands, a lowest band subband is encoded for each of the frames, and a high band subband is encoded for each of the frames. A coding method for a color still image, characterized in that the quantization values are combined into one and zero-tree coding is performed.