Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-01-27
2002-04-09
Diep, Nhon (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C348S699000
Reexamination Certificate
active
06370194
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention generally relates to a motion vector estimation circuit for estimating a motion vector for a macro block, which is a predetermined range in a frame or field of a moving picture. More specifically, the invention relates to a motion vector estimation circuit for accumulating the differences between the signal strength of the pixels of original and reference images corresponding to a motion vector candidate for the macro block, to select a motion vector candidate having a minimum accumulation result as a motion vector.
In a typical moving picture information processing, a motion vector is estimated by operating a value for a pixel in a macro block using the following technique. That is, for a certain motion vector candidate, the signal strength of a pixel in a macro block on an original image is compared with the signal strength of each of the pixels on a reference image which has a region corresponding to the macro block and which is obtained by the translating of the original image by a distance corresponding to a motion vector, to accumulate a value indicative of the difference between the signal strength for all of the pixels in the macro block. This operation is carried out for all of motion vector candidates, and the accumulated result obtained for each of the motion vector candidates is sequentially compared to estimate a candidate having a minimum accumulation result as a motion vector. In the above described technique, the term “reference image” means an image of a predetermined region corresponding to a macro block obtained by the translating of a macro block, which is a predetermined range in a frame or field of an original image, by a distance corresponding to a motion vector.
A motion vector estimation circuit for such a typical motion vector estimation is designed to carry out an accumulation processing at a high speed using a small amount of data by arranging a plurality of accumulator circuits in parallel and assigning the operation for a motion vector candidate to each of the accumulation circuits to output accumulation result data as a pipeline. In such a motion vector estimation circuit, each of the accumulation circuits is designed to carry out the operation of the signal-strength difference on the basis of the difference between the luminance value of a pixel forming a macro block of an original image and the luminance value of the corresponding pixel of a reference image. Specifically, each of the accumulator circuits carries out the operation for obtaining an absolute-value of the difference between luminance values of pixels of original and reference images, namely, |Luminance Value of Pixel of Original image—Luminance Value of Pixel of Reference Image|.
Specifically, when the motion vector estimation is carried out by the above described pipeline architecture system, the accumulation for regions corresponding to the respective motion vectors is carried out, and the same estimation is carried out for all of predetermined motion vector candidates to select and adopt a motion vector candidate, which has a minimum (or maximum) estimated value, from the motion vector candidates as a motion vector for the macro block. If the number of motion vector candidates, which can be estimated at a time, is defined as a parallel number, the motion vector estimation is completed by repeating the accumulation times obtained by dividing the number of all of the motion vector candidates by the parallel number. Therefore, when the processing is carried out by a parallel number m, each m candidates are selected from the motion vector candidates, and this processing is repeated until all of the candidates are selected. For example, if the total number of the motion vector candidates is 1024 and if the number of columns thereof is 16, 64 accumulation operations are repeated to complete the motion vector estimation.
Therefore, the motion vector estimation is not completed unless the second through number n estimations are sequentially repeated after the first estimation of the accumulated value for the motion vector candidates 1 through m is carried out. Such a motion vector estimation circuit must always carry out the accumulation operations by a fixed number of cycles regardless of the quality of an input original image, so that there are problems in that the number of cycles unnecessary for estimating motion vectors is used, and electric power consumption necessary for driving a circuit for estimating motion vectors is increased. In order to solve such problems, the inventors of the present invention have already proposed a motion vector estimation system for stopping unnecessary accumulation operation in Japanese Patent Application No. 8-341928 (1996).
This conventional motion vector estimation system comprises the construction shown in
FIG. 15
, which is provided in a motion vector estimation system for selecting a motion vector candidate of a specific region in a macro block, which is an image of a predetermined range in a frame or field of an original image, to calculate the difference between the signal strength of pixels corresponding to the motion vector candidates of original and reference images, to accumulate the signal strength difference calculated for the pixels in the macro block to derive the accumulation results, to select a motion vector candidate having a local maximum or local minimum accumulation result, as a motion vector, from the motion vector candidates in the specific region in the macro block.
That is, in
FIG. 15
, the motion vector estimation system comprises: accumulating means
1
comprising a plurality of differential absolute-value accumulating circuits
2
a
,
2
b
through
2
n
, which are connected as a pipeline for parallel inputting original image data D
1
and reference image data D
2
for motion vector candidates to carry out the accumulation of a plurality of motion vectors in parallel; accumulation stopping means
10
having a separate stop circuit
11
for stopping the accumulation operation of the accumulating means
1
for each of the accumulating circuits
2
a
,
2
b
through
2
n
; and control means
5
for controlling the supply of the original image data D
1
and the reference image data D
2
for the motion vector candidates to the accumulating circuits
2
a
,
2
b
through
2
n
of the accumulating means
1
and for outputting a separate stop signal S
3
for stopping the separate accumulation to the separate stop circuit
11
to control the stopping of the accumulation operation of the corresponding accumulating circuit
2
a
,
2
b
or
2
n.
Each of the accumulation results of the differential absolute-value accumulating circuits
2
a
,
2
b
through
2
n
forming the accumulating means
1
is outputted to a minimum-value information management circuit
4
as an accumulation signal S
1
. The minimum-value management circuit
4
compares the accumulation signal S
1
with the minimum value of the differential absolute-values which have been inputted, and outputs a comparison result signal S
2
to the control means
5
. The control means
5
outputs a stop signal S
3
to the accumulation stopping means
10
for controlling the stopping of the accumulation of each of the differential absolute-value accumulating circuits
2
a
through
2
n
forming the accumulating means
1
on the basis of the comparison result signal S
2
supplied from the minimum-value information management circuit
4
.
However, the conventional motion vector estimation system shown in
FIG. 15
is not good from the point of view of the improvement efficiency. Because if the improvement efficiency is under the influence of the quality of the input moving picture to cause the reviewing order of a motion vector candidate, which is finally to be a motion vector, to be a later order in the whole motion vector candidates, the accumulation operation for that motion vector candidate is not carried out unless the accumulation operation for almost all of motion vector candidates at the substantially front of
Diep Nhon
Hogan & Hartson L.L.P.
LandOfFree
Motion vector estimation circuit and 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 Motion vector estimation circuit and method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Motion vector estimation circuit and method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2931599