Image analysis – Image compression or coding – Including details of decompression
Reexamination Certificate
2000-02-25
2002-04-23
Grant, II, Jerome (Department: 2724)
Image analysis
Image compression or coding
Including details of decompression
Reexamination Certificate
active
06377708
ABSTRACT:
TECHNICAL FIELD
The present invention relates to an image predictive coding apparatus and method, image predictive decoding apparatus and method and recording medium. The present invention relates, in particular, to an image predictive coding apparatus and method as well as image predictive decoding apparatus and method for storing digital image data of an image which is a static image or a dynamic image into a recording medium such as an optical disk or for transmitting the data through a communication line. The present invention also relates to a recording medium in which a program including the steps of the image predictive coding method is recorded as well as a recording medium in which a program including the steps of the image predictive decoding method is recorded.
BACKGROUND ART
For the purpose of efficiently storing or transmitting a digital image, the image is required to be coded in a compression coding manner. As a method for coding a digital image in a compression coding manner, there is a waveform coding method of sub-band coding, wavelet coding, fractal coding or the like other than discrete cosine transform (referred to as a DCT transform hereinafter) represented by JPEG (Joint Photographic Experts Group) and MPEG (Motion Picture Experts Group). For the purpose of removing a redundant signal between images, an inter-image prediction with a motion compensation is executed, thereby subjecting a differential signal to waveform coding.
According to the MPEG system, an input image is processed while being divided into a plurality of 16×16 macro blocks. One macro block is further divided into 8×8 blocks and quantized after undergoing 8×8 DCT transform. This is called an intra-frame coding.
On the other hand, according to a motion detection method inclusive of block matching, a prediction macro block having the minimum error with respect to the objective macro block is detected from other frames adjacent in time, the detected prediction macro block is subtracted from the target macroblock thereby forming a differential macro block, and this macro block is quantized after undergoing 8×8 DCT transform. This is called an inter-frame coding, and the prediction macro block is called a prediction signal of the time domain.
A normal image has spatially similar regions, and an image can be approximated to a spatial region by utilizing this characteristic. In a manner similar to that of the prediction signal of the time region, a prediction signal can also be obtained from an identical frame. This is called a spatial prediction signal.
Since spatially adjacent two pixel values are close to each other, the prediction signal of the spatial region is generally located close to the target signal. On the other hand, on the receiving side or the reproducing side, a signal which has been coded and reproduced in the past is required to be used as the prediction signal since the original image is absent. From these two factors, the prediction signal of the spatial region is required to be generated at high speed. This is because the signal used for the generation of a prediction signal has to be decoded and reproduced.
Therefore, the prediction signal of the spatial region is required to be generated in a simple manner, as well as, in high accuracy. Furthermore, a quickly operable construction is required in a coding apparatus and a decoding apparatus.
The coding of image data has been widely used in many international standards such as JPEG, MPEG1, H.261, MPEG2 and H.263. Each of the latter standards has a more improved coding efficiency. That is, much effort has been devoted to further reducing the number of bits than in the conventional standards in expressing the same image quality.
Coding of image data of moving images is comprised of intra-frame coding and prediction frame coding. In a representative hybrid coding system such as MPEG1 Standard, consecutive frames can be classified into the following three different types:
(a) intra-frame (referred to as an “I-frame” hereinafter);
(b) prediction frame (referred to as a “P-frame” hereinafter); and
(c) bidirectional prediction frame (referred to as a “B-frame” hereinafter).
An I-frame is coded independently of the other frames, i.e., the I-frame is compressed without referring to the other frames. A P-frame is coded through motion detection and compensation by using the preceding frame for predicting the contents of a coded frame (it is a P-frame). A B-frame is coded through motion detection and compensation by using information from the preceding frame and information from the subsequent frame for predicting the data of the contents of the B-frame. The preceding frame and the subsequent frames could be an I-frame or a P-frame. The I-frame is coded in intra- modes. The P-frame and the B-frame are coded in intra and prediction mode.
As the characteristics of the coding of the I-frame, P-frame and B-frame are different from one another, the compressing methods thereof differ from one another. The I-frame uses no temporal prediction for the purpose of reducing the redundancy, and therefore, it requires more bits than those of the P-frame and the B-frame.
A description will be herein made taking MPEG2 as an example. It is assumed that the bit rate is 4 Mbits/sec and an image having 30 frames/sec is used. In general, the ratio of the number of bits used for the I- P- and B-frames is 6:3:1. Therefore, the I-frame uses about 420 kbits/s, and the B-frame uses about 70 kbits/s. This is because the B-frame is sufficiently predicted from both directions.
FIG. 14
is a block diagram showing a construction of a prior art image predictive coding apparatus. Since a DCT transform is executed on a block basis, the recent image coding methods are all based on the division of an image into smaller blocks. According to the intra-frame coding, an inputted digital image signal is first of all subjected to a block sampling process
1001
as shown in FIG.
14
. Next, the blocks obtained after the block sampling process
1001
are subjected to a DCT transform process
1004
and thereafter subjected to a quantizing process
1005
and a run length Huffman variable length coding (VLC: Variable Length Coding; entropy coding) process
1006
. On the other hand, according to the prediction frame coding, an inputted digital image is subjected to a motion compensating process
1003
, and the motion-compensated block (i.e., the predicted block) is subjected to the DCT transform process
1004
. Next, the quantizing process
1005
and the run length Huffman VLC coding (entropy coding) process
1006
are executed.
The fact that the block-based DCT transform process
1004
removes or reduces a spatial redundancy inside the target block to be processed and the fact that the motion detecting and compensating processes
1002
and
1003
remove or reduce a temporal redundancy between adjacent frames are known from the conventional image coding techniques. Further, the run length Huffman VLC coding or other entropy coding processes
1006
executed after the DCT transform process
1004
and the quantizing process
1005
removes statistical redundancy between quantized DCT transform coefficients. However, the process is executed only on the blocks within an image.
A digital image has a spatially great redundancy as an inherent characteristic. This redundancy exists not only in the blocks inside a frame but also between blocks over blocks. However, the fact that no actual method uses a process for removing the redundancy between blocks of an image is apparent from the above description.
According to the existing image coding method, the DCT transform process
1004
or another transform process is executed on the block basis due to restrictive conditions in terms of hardware formation and calculation.
Although the spatial redundancy is reduced through the block-based transform process, it is restricted to the inside of one block. The redundancy between adjacent two blocks is not satisfactorily considered. The redundancy, however, can be further reduced wh
Shen Sheng Mei
Tan Thiow Keng
LandOfFree
Image predictive coding method 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 predictive coding method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image predictive coding method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2868967