Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2000-02-11
2003-04-15
Rao, Andy (Department: 2713)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240170
Reexamination Certificate
active
06549576
ABSTRACT:
BACKGROUD OF THE INVENTION
1. Field of the Invention
The present invention relates to motion vector detecting method and apparatus and, more particularly, to motion vector detecting method and apparatus for processing blocks having smaller residual difference at an early stage of search.
2. Description of the Related Art
Conventionally, as a highly efficient compression coding method of an animation, H.261 and H.263 of an ITU-T (International Telecommunication Union-Telecommunication Standardization Sector) recommendation or MPEG-1 and MPEG-2 of a MPEG (Moving Picture Expert Group) recommendation in ISO (International Organization for Standardization) etc. are known. They are the highly efficient compression coding technique of time-varying images formed of combinations of techniques such as DCT (Discrete Cosine Transformation), inter-frame prediction by motion compensation, quantization, and variable length coding.
FIG. 13
is one example of a predictive coding apparatus for coding images using inter-frame prediction by motion compensation. In
FIG. 13
, image data is supplied to an input unit
100
. Also, image data (hereinafter called “reference image”) of a frame connected in time series to a current input image (hereinafter called “current image”) is stored in frame memory
102
.
A motion vector detecting unit
101
obtains a motion vector indicating correspondence every minor area of both the images using the current image supplied from the input unit
100
and the reference image supplied from the frame memory
102
. The obtained motion vector is supplied to a motion compensation unit
103
. The motion compensation unit
103
performs motion compensation processing using the motion vector supplied from the motion vector detecting unit
101
and the reference image supplied from the frame memory
102
, and obtains image data (hereinafter called “predictive image”) to which the motion compensation processing is performed.
A subtracter
104
obtains image data (hereinafter called “predictive difference image”) subtracting the predictive image from the current image using the current image supplied from the input unit
100
and the predictive image supplied from the motion compensation unit
103
.
A DCT (Discrete Cosine Transformation) unit
105
performs DCT to the predictive difference image supplied from the subtracter
104
and obtains a coefficient of DCT.
A quantizing unit
106
performs quantization processing to the coefficient of DCT supplied from the DCT unit
105
. Output of the quantizing unit
106
is outputted from an output unit
107
. Also, the output of the quantizing unit
106
is supplied to an inverse quantizing unit
108
.
The inverse quantizing unit
108
performs inverse quantization processing to the quantized coefficient of DCT supplied from the quantizing unit
106
and obtains a coefficient of DCT. An inverse DCT unit
109
performs inverse DCT processing to the coefficient of DCT supplied from the inverse quantizing unit
108
and obtains difference data from the reference image.
An adder
110
adds the difference data to the predictive image to restore the current image using the difference data supplied from the inverse DCT unit
109
and the predictive image supplied from the motion compensation unit
103
.
The frame memory
102
stores the restored current image supplied from the adder
110
.
A switching unit
111
switches whether inter-frame prediction is performed or not, and the inter-frame prediction is performed if the switching unit
111
is turned on and the inter-frame prediction is not performed if the switching unit
111
is turned off.
In the compression coding using the inter-frame prediction thus, the difference data between the current image and the predictive image obtained by performing the motion compensation processing is encoded on the basis of the motion vector detected by the motion vector detecting unit. Because of this, the higher the detecting accuracy of the motion vector detected by the motion vector detecting unit, the smaller the amount of information on the difference data, with the result that coding efficiency is improved.
FIG. 14
illustrates motion vector detection used in the compression coding using such an inter-frame prediction. In
FIG. 14
, numeral
120
designates a current image and numeral
121
designates a reference image. In the motion vector detection, an image is divided into minor areas (hereinafter called “block”) and a motion vector is obtained every block. For this purpose, a block
124
best matching a block
122
of the current image is detected within a search range
123
set in the reference image
121
to the block
122
within the current image
120
.
The difference in position within the image between the block
122
of the current image and the block
124
best matching the block
122
within the search range
123
of the reference image in this case is a motion vector
125
. As an evaluation value of matching between blocks, a total of square or absolute value of the difference of each pixel value within the block is often used. The case of using the total of absolute value of the difference of each pixel value within the block as the evaluation value of matching will be described below, and the total of absolute value of the difference of each pixel value within this block is called “residual difference”.
As a method of motion vector detection, a block matching method is known. In the block matching method, a position of blocks within the set search range is moved, and the residual difference between the blocks in each position within the search range and blocks of the current image is obtained, and the block position so as to minimize the residual difference is obtained.
In the block matching method, the problem is that the amount of operation is large. For example, if it is assumed that the size of a block is 16 pixels×16 pixels and the size of a search range is 15 pixels in longitudinal and transverse directions, respectively, and the residual difference is obtained at each integer pixel position within the search range, in order to obtain a motion vector of one block, calculations of the absolute value of 256 (=16×16) pixels of the difference have to be performed 961 (31×31) times. Further, in order to carry out this processing with respect to each block in the current image, the whole amount of operation becomes larger.
Because of this, a method of reducing the amount of operation of the block matching method is required. As the method of reducing the amount of operation, a SSDA (Sequential Similarity Detection Algorithm) method is known. This method is disclosed in, for example, DANIELI. BARNEA, HARVEY F. SILVERMAN, “A Class of Algorithms for Fast Digital Image Registration”, IEEE Transactions on Computer, Vol. C-21, No.2, pp179-186, February, 1972. This method is designed so that attention is given to the nature that the residual difference remarkably increases in blocks having a little similarity and during calculation of the residual difference, the residual difference obtained in partway is compared with a certain threshold value and if the residual difference exceeds the threshold value, the calculation of the residual difference concerning said block is aborted and thereby, the amount of operation is reduced.
In the case of using this method, it becomes a problem in setting the threshold value for determining abort of the calculation of the residual difference. If the threshold value is too large, the amount of operation becomes large since the case capable of aborting the calculation of the residual difference decreases. On the contrary, if the threshold value is too small, the global minimum value cannot be searched and thus the possibility of outputting an inadequate motion vector is increased.
As for a method of setting the threshold value, there is, for example, an automatic determination method of the threshold value disclosed in “Overlay of Image by Sequential Testing Method of Residual Difference” by Morio Onoue, Norihiko Maeda, and Mas
Dickstein Shapiro Morin & Oshinsky LLP.
NEC Corporation
Rao Andy
LandOfFree
Motion vector detecting method and apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Motion vector detecting method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Motion vector detecting method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3051301