Image analysis – Image compression or coding – Quantization
Reexamination Certificate
1998-09-28
2002-12-31
Tran, Phuoc (Department: 2621)
Image analysis
Image compression or coding
Quantization
C382S232000, C382S233000, C382S234000, C382S240000, C382S245000
Reexamination Certificate
active
06501863
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an image coding apparatus, an image coding method, an image decoding apparatus, an image decoding method and a transmission medium, applicable for a system which codes or decodes an image at a high efficiency and transmits or efficiently stores the image. More particularly, the invention relates to an image coding apparatus, an image coding method, an image decoding apparatus, an image decoding method and a transmission medium which are applicable to a compressing and extending apparatus of a high-precision fine image such as a satellite image or a medical image or a software module thereof, or a compressing and extending apparatus of texture used in games or three-dimensional CG or a software module thereof.
2. Description of the Related Art
One of the typical conventional image compressing methods is the JPEG (Joint Photographic Experts Group) method standardized by the ISO (International Organization for Standardization). This is an image compressing method using DCT (Discrete Cosine transform), in which a satisfactory coded or decoded image is available when it is possible to assign relatively numerous coding bits to each pixel.
If the number of coding bits is reduced to below a certain level, however, there occurs serious block distortion unique to DCT, leading to apparent deterioration of image subjectively. Various organizations have therefore proposed techniques for solving this block distortion unique to DCT, including, for example, a technique using DC/AC conversion known as MDCT (Modified DCT) is disclosed by J. Princen, et al. in “Subband/Transform Coding Using Filter Bank Designs Based on Time Domain Aliasing Cancellation” released in the IEEE Proceedings ICASSP 87, 50.1, pp. 216102164, 1987. According to this MDCT, block distortion becoming apparent in DVT is known to be reduce by performing transformation while overlapping a block with the surrounding blocks. In general, MDCT can transform 2M time-serial sample data into M spectra, and perfectly restore the original: time-serial sample data by means of IMDCT (inverse MDCT).
As conventional arts concerning MDCT, these are available a “signal transforming apparatus” disclosed in Japanese Patent Publications Nos. 6-66067 and 06-66068, and “a normal transform calculating apparatus and an inverse transform calculating apparatus for improved DCT” disclosed in Japanese Unexamined Patent Publication No. 4-44099. These inventions relate to a method for reducing the hardware volume of MDCT and a high-speed calculation of MDCT.
In an “image coding apparatus” disclosed in Japanese Unexamined Patent Publication No. 7-50835, on the other hand, it is taught that a dynamic image can be coded by splitting the image into subbands, and applying quadrature transformation to the thus split low-level components.
Unlike the reduction of the hardware volume of MDCT or achievement of a higher speed of calculation, the objects of the Japanese patent Publications Nos. 6-766067 and 6-66068 and Japanese Unexamined Patent Publication No. 4-44099, the present invention has an object to accomplish high-efficiency image coding and decoding by the concrete application of MDCT as a technique for quadrature transformation. Therefore, Japanese Unexamined Patent Publication No. 7-50835 in which quadrature transformation is actually applied and subband splitting is carried out in the first stage of the process will be described as an example of the conventional arts.
FIG. 18
is a block diagram illustrating a configuration of a first embodiment of the aforementioned Japanese Unexamined Patent Publication No. 7-50835.
In
FIG. 18
, a subband splitter
200
can subband-split an original image into a low-level image and a plurality of high-level images. A quadrature transformer
201
applies quadrature transformation by splitting a low-level image
301
provided as an output from the subband splitter
200
into blocks of a first size, and outputs the same as blocks
302
. A block splitter
202
splits a plurality of high-level images generated by the subband splitter
200
and supplied via the quadrature transformer
201
into blocks of a first size, and can perform hybrid transformation by synthesizing the blocks
302
of a first size quadrature-transformed by the quadrature transformer
201
and blocks of a first size of high-level images to generate a block of a second size.
The subband splitter
200
, the quadrature transformer
201
and the block splitter
202
constitute a hybrid transformer
213
.
A mode selector
203
compares a hybrid transformation coefficient
303
output from the block splitter
202
and a differential transformation coefficient
304
output from an adder
204
, and changes over between a switch
215
and a switch
216
.
The adder
204
calculates a difference between, the hybrid transformation coefficient
303
of an original image output from the block splitter
202
and a hybrid transformation coefficient
305
of a reference image output from a motion compensation predictor
205
.
The motion compensation predictor
205
predicts a motion compensation of the hybrid transformation coefficient
303
. A motion vector detector
206
detects a motion vector by comparing an output of a subband synthesizer
212
and an original image
300
.
A quantizer
207
quantizes a signal from the switch
215
and outputs the same. An inverse quantizer
208
inverse-quantizes a signal
310
quantized by the quantizer
207
.
An adder
209
adds and outputs an output signal
311
from the inverse quantizer
208
and a signal
307
from the switch
216
.
A frame memory
210
rearranges the hybrid transformation coefficients
303
to generate a low-level image and a plurality of high-level images. An inverse quadrature transformer
211
splits the low-level image into blocks of a first size to conduct inverse quadrature transformation. The subband synthesizer
212
form the entire image through synthesis of subbands.
The frame memory
210
, the inverse quadrature transformer
211
and the subband synthesizer
212
constitute a hybrid inverse transformer
214
.
Operations of the aforementioned conventional art will now be described.
The subband splitter
200
subband-splits an original image into a low-level image
301
and a plurality of high-level images, and the quadrature transformer
201
splits the low-level image
301
into blocks of a first size to subject them to quadrature transformation. The block splitter
202
splits the plurality of high-level images further into blocks of the first size. It further synthesizes the quadrature-transformed blocks
302
of the first size and the blocks of the first size of the high-level images to form blocks of a second size, thereby performing a hybrid transformation. These are the operations in the hybrid transformer
213
.
The frame memory
210
prepares a low-level image and a plurality of high-level images b rearranging the hybrid transformation coefficients
312
. The inverse quadrature transformer
211
splits the low-level image into blocks of the first size and subjects the split blocks to an inverse quadrature transformation. The subband synthesizer
212
subband-synthesizes the entire image. These are the operations of the hybrid inverse transformer
214
. As a result of these operations for hybrid inverse transformation, a decoded image
318
is generated. The resultant decoded image
318
is entered into the motion compensation predictor
205
, where motion compensation prediction of the hybrid transformation coefficient
303
is conducted.
The adder
204
calculates a difference between the hybrid transformation coefficient
303
of the original image output from the block splitter
202
and the hybrid transformation coefficient
305
of the reference image output from the motion compensation predictor
205
.
The mode selector
203
compares the hybrid transformation coefficient
303
output from the block splitter
202
and the differential transformation coefficient
304
output from the adder
204
Akagiri Kenzo
Fukuhara Takahiro
Kimura Seiji
Bayat Ali
Sonneschein Nath & Rosenthal
Sony Corporation
Tran Phuoc
LandOfFree
Image coding apparatus, image coding method, image decoding... 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 coding apparatus, image coding method, image decoding..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image coding apparatus, image coding method, image decoding... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2956299