Image analysis – Image compression or coding – Pyramid – hierarchy – or tree structure
Reexamination Certificate
1998-08-04
2003-09-02
Tran, Phuoc (Department: 2621)
Image analysis
Image compression or coding
Pyramid, hierarchy, or tree structure
C382S233000, C382S244000, C382S173000
Reexamination Certificate
active
06614939
ABSTRACT:
BACKGROUND OF THE INVENTION
(1) Field of the Invention
The present invention relates to image compression encoding and decoding.
(2) Related Art
Advances in network technology and multimedia applications in recent years have created a greater need to handle large amounts of data, such as image files.
Image files generally include a large amount of data, and effectively need to be compressed before transmission. To address this need, image compression methods which use region division (hereinafter “segmentation”) have been developed.
Image compression methods that use segmentation use a predetermined judgement standard (called a “division judgement condition”) to divide image regions into segments.
As an example method of processing an image region, Image Coding by
Adaptive Tree
-
Structured Segmentation
, IEEE, Trans. Info., Vol.38, No.6, pp.1755-1767 (1992) teaches a method for dividing a region into convex polygons, while
Image Compression via Improved Quad
-
tree Decomposition Algorithms
, IEEE Image. Proc., Vol.3, No.2, pp.207-215, (1994) teaches a method for quadtree division.
FIG. 1
is a representation of quadtree division.
As shown in
FIG. 1
, quadtree division uses a predetermined division judgement condition to judge whether a square region should be divided. On judging that division is necessary, this method divides the square region into four square subregions The method then repeats the judgement for the resulting subregions. In this way, the method generates subregions called “segments”.
In the example shown in
FIG. 1
, the first division generates four segments. Of these, three segments are subjected to a second division. Of the segments produced by the second division, only segments that conform to the division judgement condition are subjected to a third division.
One conventional example of the division judgement condition uses the distribution of luminance values in the segment. Note that in this specification, “luminance” also includes the concept of “density”. This method may judge whether a difference between the highest and lowest luminance values is within a given threshold, and has an advantage in that it uses little calculation. Another conventional method uses a calculated amount of error (such as the difference of squares) for when a mean value is used in place of every element in the segment. This second method has a different advantage, however, in that users can control the quality of the final reproduced image using the information-to-noise ratio when quadtree division is performed, the compressed image data is composed of segmentation information and luminance information. For the example shown in
FIG. 1
, the segmentation information is “1,1110,0011,0010,0110”, where “1” is the division code and “0” the no-division code. The luminance information, meanwhile, 1:1 corresponds to each segment when division is complete.
When using the quadtree division method, the greater the extent to which large regions are left undivided, the smaller the data size of the compressed image data, or in other words, the higher the compression rate.
In fields such as medicine, however, images need to be reproduced with a very high resolution to allow accurate representation of detail. As a result, there is a very strong demand in lossless or near-lossless image compression. Lossless compression refers to reversible compression where the pre-compression image will be faithfully reproduced after decoding with no loss of image quality.
One example of a reversible image compression technique that uses segmentation is taught by
MDL Genri to
2
Bunkikouzou Segumenteeshon wo Mochiita Gazou no Mubizumi Fugouka Arugorizumu
(“Image Encoding Algorithm that uses MDL Principles and a Dual-Tree Structure”) published in Shin Gakuron (D-II), Vol.J80-D-II, No.2, pp.415-425, 1997).
When compressing an image using segmentation, there is the problem that improving the detail in the reproduction image inevitably results in division into segments of large areas in the original image where there are only minute differences in luminance values, or in other words, large areas may be regarded having a uniform luminance value. This significantly reduces the compression rate, and is major drawback for reversible (lossless) and quasi-reversible (near-lossless) compression.
The following is a description of the idea of “progressiveness”.
Progressive reproduction refers to a process whereby a low-resolution image is initially displayed at the start of input or decoding of a bitstream, with the details of the image being slowed added as the input or decoding progresses. The reproduction image is finally displayed with the desired resolution.
This kind of reproduction can be highly effective, and is particularly valuable when searching through images, for example.
Compressed image data obtained through an image compression technique using segmentation where each segment has its own luminance value is unsuited to progressive reproduction. When such data is decoded, even if it were possible to produce the reproduction image in accordance with gradual changes in luminance values, it would still be difficult to perform progressive reproduction in accordance with the changes in resolution and in the shapes of segments.
In view of the stated problems, it is necessary to provide a decoding apparatus with a separate component to enable the progressive reproduction of image data that has been compressed in this way. In the third document cited above, progressive reproduction is realized by using a 2-path system where a differential image for differences between the original image and the resemblance image is encoded separately to the resemblance image.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide an image compression apparatus that performs reversible compression without being limited to a low compression rate of data as in the conventional art. Such image compression apparatus generates compressed image information that is suited to progressive image reproduction. At the same time, the present invention aims to provide an image decoding apparatus that decodes compressed image data generated by this kind of image compression apparatus.
To achieve the stated object, the image compression apparatus of the present invention performs image compression using a first segmentation method on a first plane set that is composed of at least one consecutive bit plane in block image information in the original image information. By doing so, the image compression apparatus generates first region information composed of first segmentation information showing a segmentation result for a block, and first luminance information showing luminance information for each segment produced by the segmentation. The present image compression apparatus also performs image compression using a second segmentation method on a second plane set that is composed of at least one consecutive bit plane adjacent to the first plane set at a lower bit position in the block image information. By doing so, the image compression apparatus generates second region information composed of second segmentation information showing a segmentation result for the block, and second luminance information showing luminance information for each segment produced by the segmentation. The present image compression apparatus then generates compressed image information based on the first region information and second region information and outputs the compressed image information.
Note that the expression “bit plane” refers to a level (plane) composed of luminance bit values at a corresponding bit position in the luminance values.
The present image decoding apparatus reads first segmentation information and first luminance information from the inputted compressed image information, uses the first segmentation information to perform segmentation of the block according to a first segmentation method, and assigns the first luminance information to the segments produced by the segmentation to decode the first plane set.
The presen
Matsushita Electric Industrial Co. LTD
Tran Phuoc
LandOfFree
Image compression apparatus and decoding apparatus suited to... 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 compression apparatus and decoding apparatus suited to..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image compression apparatus and decoding apparatus suited to... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3080931