Image analysis – Image compression or coding – Contour or chain coding
Reexamination Certificate
1997-09-17
2001-06-26
Chang, Jon (Department: 2723)
Image analysis
Image compression or coding
Contour or chain coding
C382S197000, C382S199000
Reexamination Certificate
active
06252993
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a video signal coding method and a decoding method, and more particularly, applicable is to the case where feature points of an image are detected and then high-efficiency coding of the video signal is performed.
2. Description of the Related Art
Heretofore, as a method for high-efficiency coding of a video signal, there has been a method wherein an input video signal is serial-parallel converted by means of a DCT (discrete cosine transform), and then adaptive quantization according to a human visual property is performed for each frequency band, or a method wherein an image is subband-divided using wavelet base and then weighted so as to code for each band. According to these with methods, distortion is not particularly visible and a high compression rate can be obtained.
However, in these coding methods, such a disadvantage exists that a visual undesirable effect such as a block distortion becomes noticeable as a compression rate is raised. Then, as a coding method which can suppress a visual undesirable distortion under a high compression rate, there is a structure extracting and coding method according to change point detection of an image, in which the distinctive points of the structure of the image are extracted and effectively coded.
For example, as shown in
FIG. 1
, in a structure extracting and coding apparatus
1
which codes a video signal using a structure extracting and coding method, an input video signal S
1
is quantized by a quantizer
2
, thereby a quantization coefficient S
2
is generated, which is sent to a multiplexer
3
.
On the other hand, the structure extracting and coding apparatus
1
inputs the input video signal S
1
to a two-dimensional change point detecting circuit
4
, which detects the feature points, and sends a flag “1” to a chain coder
5
as a feature point signal S
3
if the present signal is detected as the feature point, otherwise sends a flag 0. A chain coder
5
chain codes the pixels of which feature point signals S
3
are “1” in one frame, and sends it to the multiplexer
3
as a chain code output S
4
.
The multiplexer
3
multiplexes the chain code output S
4
and the quantization coefficient S
2
at the feature point, and then sends this to a buffer memory
6
as a multiplexed feature point signal S
5
. The buffer memory
6
smooths an information quantity of the multiplexed feature point signal S
5
, and outputs it as an output signal S
6
of the structure extracting and coding apparatus
1
according to feature point detection of the image.
Here, the two-dimensional change point detecting circuit
4
is constructed, for instance, as shown in FIG.
2
. More specifically, the two-dimensional change point detecting circuit
4
once stores the input video signal S
1
in a frame buffer
10
, and then smooths it by means of a filter
11
. A smoothed signal S
10
output from the filter
11
is sent to a lateral direction processing block
12
and a longitudinal direction processing block
13
, respectively.
The lateral direction processing block
12
supplies the smoothed signal S
10
to an adder
16
via a delay element
14
and a multiplier
15
of which multiplication coefficient is −1 respectively, thereby a difference between the present signal and the signal preceding one sample is found by the adder
16
, so that a lateral direction amplitude inclination signal S
11
is generated. The lateral direction amplitude inclination signal S
11
is squared by a multiplier
17
, thereby a lateral direction amplitude inclination signal power S
12
is obtained. The lateral direction amplitude inclination signal power S
12
is supplied to a multiplier
21
via a delay element
18
and an adder
19
to which an inverted threshold value T is given, and also supplied to the multiplier
21
via an adder
20
to which the inverted threshold value T is given. Thereby, in the multiplier
21
, a difference signal S
13
between the lateral direction amplitude inclination signal electric power S
12
and the threshold value T is multiplied by a difference signal S
14
between the one sample preceding lateral direction amplitude inclination signal power S
12
and the threshold value T, and the multiplication result signal S
15
is sent to a comparator
22
. The comparator
22
sends a flag “1” in the case where the multiplication result signal S
15
is less than 0, or a flag “0” in the case where it is “0” or more to an OR circuit
23
as a lateral direction change point detection signal S
16
.
The longitudinal direction processing block
13
supplies the smoothed signal S
10
to an adder
26
via a line delay
24
and a multiplier
25
respectively, thereby a difference between the present signal and the signal preceding one line is obtained by the adder
26
, so that a longitudinal direction amplitude inclination signal S
17
is generated. The longitudinal direction amplitude inclination signal S
17
is squared by a multiplier
27
, thereby a longitudinal direction amplitude inclination signal power S
18
is obtained. The longitudinal direction amplitude inclination signal power S
18
is supplied to a multiplier
31
via a line delay
28
and an adder
29
to which an inverted threshold value T is given, and also supplied to the multiplier
31
via an adder
30
to which the inverted threshold value T is given. Thereby, in the multiplier
31
, a difference signal S
19
between the longitudinal direction amplitude inclination signal power S
18
and the threshold value T is multiplied by a difference signal S
20
between the longitudinal direction amplitude inclination signal power S
18
preceding one line and the threshold value T, and the multiplication result signal S
21
is sent to a comparator
32
. The comparator
32
sends a flag “1” in the case where the multiplication result signal S
21
is less than “0” or a flag “0” in the case where it is “0” or more to the OR circuit
23
as a longitudinal direction change point detection signal S
22
.
By ORing the lateral direction change point detection signal S
16
with the longitudinal direction change point detection signal S
22
, the OR circuit
23
outputs a two-dimensional change point detecting circuit output signal S
23
.
In coding of feature point coordinates based on the conventional two-dimensional change point detecting method and chain coding described above, with respect to the change points of the linear differential value power of the input video signal against a certain threshold value, for instance the points at which a sharp change arises locally, the two-dimensional change point detecting circuit
4
at first detects two points of the starting point and the ending point of that change as feature points, and the following chain coder
5
separately chain codes.
Therefore, in the conventional feature point coordinates coding method, on the edge portion at which a sharp change arises, the position coordinates of the two curves are always coded, hence such a disadvantage has existed that the code quantity increases correspondingly.
SUMMARY OF THE INVENTION
In view of the foregoing, an object of the present invention is to provide a video signal coding method and a decoding method which are possible to decrease the total code quantity by decreasing the bit quantity required for coding of the coordinate information when the feature point of the input video signal is detected and then the coordinate of the feature point is chain coded.
The foregoing object and other objects of the invention have been achieved by the provision of a video signal coding method for detecting feature points of an input video signal (S
1
) and for coding coordinates of the feature points, comprising the steps of: detecting feature points (S
30
) of the input video signal (S
1
) by canny edge detecting method; performing chain coding (S
33
) of the coordinates (S
30
) of the feature points detected at the feature point detecting step; coding the length of the region (S
32
), including the feature points, of which edge intensity
Chang Jon
Sonnenschein Nath & Rosenthal
Sony Corporation
LandOfFree
Video signal coding and decoding 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 Video signal coding and decoding method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video signal coding and decoding method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2539909