Image analysis – Image compression or coding – Adaptive coding
Reexamination Certificate
1998-12-14
2001-05-22
Couso, Jose L. (Department: 2724)
Image analysis
Image compression or coding
Adaptive coding
Reexamination Certificate
active
06236756
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a device and method for analyzing an input image and calculating an image coding parameter for compressing the image.
2. Discussion of the Related Art
Recently, a capacity of a storage medium or transmission time has been reduced by compressing images. The images electronically stored or transmitted come to be of high resolution and full-colored, and thus of a large capacity. Therefore, it is important to raise a ratio of image compression. Hereinafter, image coding and image compression indicate the same meaning.
There are two types of image compression methods: the lossless image compression and the lossy image compression. When the lossless image compression method is used, a compressed image is completely restored to what it was by decompressing. When the lossy image compression method is used, a compression ratio is expected to be higher than that of the lossless image compression method. However, a compressed image cannot be completely restored by decompressing and the image quality is deteriorated.
In general, under the same coding conditions, the compression ratio and the image quality are inversely proportional to each other. In other words, when the compression ratio is low, a small amount of information is lost and the image quality is maintained. However, if the compression ratio is raised, a large amount of information is lost and the image quality is deteriorated.
An image compression apparatus or method can control the image compression ratio and the image quality by changing an image coding parameter.
In compressing an image, it is desired to maintain a predetermined image quality while compressing the image with the compression ratio as high as possible. That is, it is required to control the coding parameter to provide the best tradeoff between the compression ratio and the image quality so that the compressed image is visually lossless in comparison with the image before compression.
In many cases, an input image has local image characteristics different from one another. In some portions, deterioration in image quality is easily observed and in the other portions, deterioration in image quality is hardly observed. Considering the image quality of the input image as a whole, it is necessary to lower the overall compression ratio to reduce deterioration in image quality in the portions where the deterioration is easily observed.
The following method makes it possible to raise the overall compression ratio while maintaining the image quality.
An input image is divided into blocks. It is then determined whether the deterioration in the image quality is easily observed in each of the divided blocks. A coding parameter that lowers the compression ratio is provided to the blocks where the deterioration in the image quality is easily observed. On the other hand, a coding parameter that raises the compression ratio is provided to the blocks where the deterioration of the image quality is not easily observed. Thereby, the compression ratio in the portions of the image where the deterioration in the image quality is rarely observed can be raised. Thus, the overall compression ratio can be raised while the image quality is uniformly maintained.
In this method, the different coding parameter is selected for each of the blocks. Therefore, the selected coding parameter is added to the head of the code of each block as shown in FIG.
14
and is also coded.
FIG. 19
shows a configuration example of an image coding apparatus used in general to which the present invention is to be applied. In the figure, an input image
1901
is divided by an input image dividing circuit
1902
. A divided image
1907
is transmitted to an image quality analysis circuit
1903
and a lossy coding circuit
1905
. The image quality analysis circuit
1903
analyzes the divided image
1907
and outputs a coding parameter
1904
. The lossy coding circuit
1905
performs coding of the divided image
1907
using the coding parameter
1904
and outputs code
1906
.
The present invention relates to the image quality analysis circuit
1903
in FIG.
19
.
Conventional examples are described more specifically. As a concrete method able to control the image quality or compression ratio by changing the coding parameter, JPEG is described, for example, in “International Standard of Multimedia Coding”, Maruzen Publishing Company, pp. 18-43. In a method using Discrete Cosine Transform (DCT) represented by JPEG, a higher compression ratio is available with the same image quality by using a quantization matrix adaptive to the characteristics of input image blocks.
The DCT is briefly explained by reference to FIG.
15
.
In the figure, input image information
1501
is divided into rectangular blocks by a blocking circuit
1502
. An orthogonal transform circuit
1503
performs an orthogonal transform on each of the divided blocks of the image information and outputs an orthogonal transform coefficient
1504
. The orthogonal transform coefficient
1504
is quantized with a predetermined quantization matrix by a quantization circuit
1505
. A coding circuit
1506
provides a code to the quantized orthogonal transform coefficient and outputs it as a code
1507
.
In the coding method as shown in
FIG. 15
, the same quantization process is performed on all blocks. Therefore, deterioration in image quality is caused in the blocks where distortion is prone to occur. On the other hand, in the blocks where distortion rarely occurs, visually useless and unnecessary information is also coded.
As aforementioned, there is a method able to raise the compression ratio while maintaining the image quality by performing different quantization processes on image portions where deterioration in image quality is easily observed and on those where deterioration is rarely observed. The method is explained by reference to FIG.
16
. Constituents corresponding to constituents in
FIG. 16
have the same reference numbers as those in FIG.
15
.
In
FIG. 16
, each of the divided blocks of the image information is transmitted to an image analysis circuit
1508
and characteristics of the image in the block are analyzed. A result of the analysis
1509
is transmitted to a quantization process selection circuit
1510
and a quantization process is selected. In the DCT coding method, specifically, the quantization process selection circuit
1510
selects a quantization matrix optimum to the result of the analysis
1509
. The selected quantization process
1511
is transmitted to the quantization circuit
1505
to be performed. Other circuits operate as same as those corresponding thereto shown in FIG.
15
.
In the above-described methods, the image analysis method for analyzing an image and calculating a coding parameter is important for improving both image quality and compression ratio. There are many types of analysis methods. Two examples of conventional image analysis device are described as follows by reference to FIG.
16
.
FIG. 16
shows main parts extracted from the two conventional examples and rearranged.
A conventional example 1 is disclosed by Japanese Patent Application Laid-Open No. 6-165149. It is determined whether input image blocks are suitable to be coded by the coding method employed here. If they are determined to be suitable, high image quality is expected in the result of coding, and accordingly, they are coded with a high compression ratio. Otherwise, if they are not suitable, low image quality is expected. Therefore, the blocks are coded with a low compression ratio to improve the image quality.
In the conventional example 1, the image analysis circuit
1508
shown in
FIG. 16
calculates a physical quantity
1509
indicating probability of occurrence of mosquito noise in each block. The quantization process selection circuit
1510
selects a quantization matrix corresponding to the physical quantity
1509
to raise the compression ratio while maintaining the same image quality. Then a code amount in the blocks showing high probabili
Kamizawa Koh
Kimura Shun-ichi
Koshi Yutaka
Kunitake Setsu
Couso Jose L.
Do Anh Hong
Fuji 'Xerox Co., Ltd.
Oliff & Berridge PLC.
LandOfFree
Image analysis method and device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image analysis method and device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image analysis method and device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2558696