Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-01-20
2002-09-24
Kelley, Chris (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C382S236000, C348S416100
Reexamination Certificate
active
06456660
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a device and method of detecting motion vectors. More particularly, the present invention relates to a technique of decreasing computational complexity for motion vector detection and realizing proper motion vector detection.
The motion compensated inter-frame prediction system using block matching operations has been standardized and widely used as a technique of efficiently encoding motion image data. The interframe prediction requires to detect motion vectors. In order to capture large motion vectors, the vector value of a motion vector must be widely searched. However, a wide search range of motion vectors leads to the enormous computation for the matching process, thus causing the disadvantage in enlarging the system scale and in making the real-time processing difficult.
In order to deal with such problems, JP-A No. 166684/1989 (hereinafter referred to as a prior art 1) discloses a motion vector detection device while JP-A No. 328333/1993 (hereinafter referred to as a prior art 2) discloses a motion vector detection device.
FIG. 19
is a block diagram illustrating the configuration of the motion vector detecting device according to the prior art
1
. In the motion vector detection device, the frame memory
201
stores reference image data. The reference point setting section
202
sets reference points referred to according to the search motion vector by using the reference image data stored in the frame memory
201
based on an offset set by the horizontal offset control section
206
and an offset set by the vertical offset control section
207
. The subtracting section
203
obtains a difference between input data and a reference point. The frame difference addition section
204
sums differential data sets being output results from the differential section
203
. The minimum value detection section
205
detects a minimum value of the sum and then detects a motion vector to provide a horizontal motion vector component Vx and a vertical motion vector component Vy. The adding section
210
adds an offset value selected by the selector
208
to the motion vector horizontal component Vx from the minimum value detection section
205
. The adding section
211
adds an offset value selected by the selector
209
to the motion vector horizontal component Vy from the minimum value detection section
205
. Thus, the final motion vector (horizontal motion vector component and vertical motion vector component) is computed. The horizontal offset control section
206
sets an offset value for motion vector search in the next frame by the horizontal component Vx of a motion vector from the minimum value detection section
205
. The vertical offset control section
207
sets an offset value for motion vector search in the next frame by the vertical component Vy of a motion vector from the minimum value detection section
205
.
In the offset setting operation of the horizontal offset control section
206
and the vertical offset control section
207
, one motion vector component is expressed by the following formula:
&dgr;
V
(
n
+1)=&dgr;
V
(
n
)+(&agr;×&Dgr;
V
)
where &dgr;V(n) is the offset value of the n-th frame; and &Dgr;V is a search range. One component V of a motion vector is expressed by the range satisfying −V≦V≦&Dgr;V. The value of &agr; is expressed the formula of &agr;=1 (V=&Dgr;V), &agr;=−1 (V=&Dgr;V), or &agr;=0 (V≠&Dgr;, −&Dgr;V)).
FIG. 20
is a block diagram illustrating the configuration of the motion vector detecting device according to the prior art
2
. In the motion vector detection device, the motion vector detection circuit
301
detects a motion vector providing a minimum prediction error based on image data for an input current frame and reference image data over a search range of specified by the selector
304
. The motion vector memory
302
stores the detected motion vector for the duration of one frame and then outputs it as a motion vector at the same position on the previous frame to the search range decision section
303
.
The search range decision section
303
judges the search range of motion vectors in a target block within the current frame according to the magnitude and direction of the motion vector of the previous frame. The selector
304
selectively outputs one of N search ranges ranging from the first search range to the n-th search range according to decision results of the search range decision section
303
. The motion vector detection circuit
301
searches motion vectors over a selected search range.
In addition to the normal search range shown in FIG.
21
(
a
), the selector
304
outputs the search range attaching importance to the horizontal direction as shown in FIG.
21
(
b
), the search range attaching importance to the vertical direction as shown in FIG.
21
(
c
), or the search range attaching importance to the slanting direction as shown in FIG.
21
(
d
). The search range attaching importance to the horizontal direction as shown in FIG.
22
(
b
) and the search range attaching importance to the vertical direction as shown in FIG.
22
(
c
) can be used as a modified search range, in addition to the normal search range shown in FIG.
22
(
a
)
However, the prior arts 1 and 2 have the following drawbacks:
In the prior art 1, the search range cannot be followed under an occurrence condition of some motion vectors so that the optimum search range of motion vectors cannot be covered. As a consequence, the prior art 1 cannot suitably detect motion vectors.
The prior art 2 has a disadvantage in that the motion vector cannot be suitably detected because only limited types of search range can be selected. Moreover, since the search range over a motion boundary portion is often predicted erroneously, the motion vector cannot be suitably detected.
As a related art, JP-A-No. 30407/1993 (hereinafter referred to as prior art 3) discloses a motion vector detection technique for realizing the automatic mechanism such as autofocus, autoiris or the like for cameras.
In the prior art 3, a partial physical region in the image pickup region
401
acts as the motion vector detection region
403
for motion vector detection, as shown in FIG.
23
.
The motion vector detection region
403
is divided into plural sub-divided regions
404
to create a histogram showing vector values of motion vectors detected for the sub-divided regions
404
. The motion vector detection region
403
on the next frame is decided by setting the motion vector with the highest occurrence frequency on the histogram as the motion vector of the subject
402
.
However, the process ranging from creation of a histogram to the computation of the motion vector detection region
403
on the next frame is realized by software. Hence, where the method is applied to encode motion images needed to detect motion vectors of the whole image region within one frame, a sophisticated processing circuit such as processor or microprocessor with considerable high processing capability is required.
When an image contains a still portion, the occurrence frequency of (0, 0)in a detected motion vector becomes high. However, in the prior art 3, since a correction process is not performed to motion vectors with high occurrence frequency at (0, 0), the motion vector detection region
403
may not be suitably set according to the motion of the subject
402
.
Moreover, since the same matching process must be performed for motion vector detection, regardless the degree of the motion of the subject
402
, the computational complexity for the matching process becomes enormous to capture a specially large motion vector.
SUMMARY OF THE INVENTION
The present invention is made to solve the above-mentioned problems. The objective of the present invention is to provide a motion vector detection device that can reduce the computational complexity by limiting the range searching for the vector values of a motion vector for, particularly, motion image encoding application
NEC Corporation
Philippe Gims
Scully Scott Murphy & Presser
LandOfFree
Device and method of detecting motion vectors does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Device and method of detecting motion vectors, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Device and method of detecting motion vectors will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2861878