Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2000-12-26
2003-06-24
Diep, Nhon (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240160, C382S236000, C348S699000
Reexamination Certificate
active
06584155
ABSTRACT:
CROSS REFERENCE TO RELATED APPLICATION
This application claims benefit of priority under 35 U.S.C. §119 to Japanese Patent Application No. 1999-369353, filed on Dec. 27, 1999, and Japanese Patent Application No. 2000-387684, filed on Dec. 20, 2000, the entire contents of which are incorporated by reference therein.
BACKGROUND OF THE INVENTION
The present invention relates generally to a method and system for estimating a motion vector. More specifically, the invention relates to a motion vector estimating method and system capable of searching a candidate motion vector of a picture element (pixel), which projects from a reference screen, with a precision of, e.g., half pixels, in a process for estimating a motion vector of a dynamic image, and a recording medium in which a motion vector estimating procedure has been recorded.
In general, a motion vector estimating system is used for compensating the motion of dynamic images and for compressing dynamic images. As systems which are often used for compressing dynamic images, there are MPEG1 and MPEG2 which are international standards having been standardized by MPEG (Moving Picture Experts Group). In MPEGs serving as these standards, a motion-compensated coding technique is used between temporally spaced screens in order to compress redundancy in a time direction. In this case, it is required to estimate motion vectors.
A typical motion vector estimation uses a technique for carrying out a block matching between a coded block on a coded image and a plurality of candidate blocks, which exist in a predetermined search area on a reference image and which have the same size as that of the coded block, and for determining a correlation between the blocks to use a displacement of a block having the highest correlation from the coded block, as a motion vector. The coded block means a block on the current image which is to be coded. The reference screen means the whole screen including a reference image. The candidate block means a block on a reference image to which reference is made when a motion vector of a certain coded block is estimated. The search area means a certain area in a reference screen including all of candidate blocks for a certain coded block.
In the block matching, the absolute value of the difference between pixels arranged at the same positions in two blocks is often calculated to be added with respect to pixels in the whole area of the blocks to be used as an index value indicative of the correlation. In this case, the correlation is high as the index value is small.
In a test model of MPEG2, it is recommended to use a precision of one pixel in the primary search and a precision of half pixels in the secondary search, as a precision of search of a motion vector. Usually, the original image is used as a reference image in the primary search, and a local decoded image, which is obtained by decoding image data coded in an encoder, is used as a reference image in the secondary search.
With respect to the processing of the edge of a reference screen during a motion vector estimation with a precision of one pixel, in a well-known motion vector estimating system, e.g., in a MPEG2 coding LSI (Large Scale Integrated-circuit) produced as a motion vector estimating system, a correction is carried out so that a search area does not project from the reference screen, i.e., an obtained motion vector is located within the reference screen, when a motion vector is estimated with a precision of one pixel in a certain coded block.
However, if a motion vector with a precision of one pixel in a certain coded block indicates the boundary of a reference screen, when a motion vector is estimated with a precision of half pixels using the motion vector with a precision of one pixel as the center of search, there is the possibility that a part of pixels in one or more candidate blocks in the search area project outwardly from the reference screen. For example, in
FIG. 1
schematically showing the vicinity of the boundary of the reference screen, with respect to the uppermost leftmost one of candidate motion vectors with a precision of half pixels in an area
203
, candidate blocks including the uppermost leftmost one as an upper left pixel correspond to an area, which is surrounded by a chain line in the figure, to include pixels outside of the reference screen.
In
FIG. 1
, a motion vector
201
with a precision of one pixel is position at the upper left corner of a screen edge of the reference screen. An area
204
projecting from the screen edge
202
of the reference screen exists in the area
203
of candidate motion vectors with a precision of half pixels in a case where the motion vector
201
is used as the center of search, and the candidate motion vectors in the area
204
can not be used for the operation of a motion vector estimation, so that only points in an area
205
among the candidate motion vectors with a precision of half pixels are actually used as candidate motion vectors.
The above described conventional coding LSI copes with such a case by carrying out the processing for estimating a candidate block including projecting pixels and excluding the estimated candidate block from the candidates for a motion vector estimation with a precision of half pixels.
In an image coding system disclosed in Japanese Patent Laid-Open No. 5-227525 filed by the applicant, there is provided means for controlling at least one of a search area in a secondary motion search and a motion vector estimation precision, in accordance with an area from which a motion vector estimated by a primary motion search is estimated in the search area, particularly as set forth in claim 5.
Also in the conventional motion vector estimating system, when the motion vector with a precision of one pixel indicates the boundary of the reference screen and when a part of the candidate blocks during the motion vector estimation with a precision of half pixels corresponding thereto project from the reference screen, the candidate blocks required for the motion vector estimation with a precision of half pixels can be searched by the above described two methods.
However, in the former prior art, the number of candidate motion vectors is smaller than that when all of the search area is located within the reference screen. For example, only four candidate motion vectors shown in the area
205
of the candidate motion vectors with a precision of half pixels in the area
203
shown in
FIG. 3
are used as candidate motion vectors. Also when candidate blocks having the highest correlation exist just nearby the original search area on the reference screen, the candidate blocks are not located within objects to be estimated, so that there is a problem in that the estimation precision deteriorates in the motion vector estimation with a precision of half pixels obtained by the secondary motion estimation.
In the latter prior art, positions in the search area for primary motion vectors are used for controlling the search area during the secondary motion search, so that there is a problem in that it is not possible to determine whether the primary motion vector is positioned at the edge of the reference screen if position information in the above described search area on the reference screen is not obtained. In addition, when a hardware resource is used for calculating the correlation between coded blocks and candidate blocks, the construction of candidate blocks outwardly projecting from the reference screen is wasted, so that there is a problem in that the use efficiency of the hardware resource deteriorates.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to eliminate the aforementioned problems and to provide a motion vector estimating system and method capable of putting a search area, which is used when a motion vector estimation with a precision of half pixels is carried out, in a reference screen by adding a correction value when a motion vector of a pixel located at the edge of a reference screen is estimated, thereby improving the precision
Takeda Hideki
Yahagi Kunihiko
Diep Nhon
Kabushiki Kaisha Toshiba
LandOfFree
Method and system for estimating motion vector does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for estimating motion vector, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for estimating motion vector will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3159373