Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2000-12-06
2004-05-18
Philippe, Gims (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C382S236000, C348S699000
Reexamination Certificate
active
06738426
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus and method for detecting motion vector suited for use in high-efficiency coding of video data.
2. Description of the Related Art
For motion prediction in TM5 (test mode 5) of the international standard ITU-T H.262/MPEG2 for video data compression, the motion is fully examined on the basis of half of pixel. However, this system requires the handling of a large amount of data. Therefore, a motion estimating process is divided into two steps: a step of calculating a motion vector in two-pixel precision using a first stage motion estimator and a step of calculating a motion vector in half-pixel precision from the motion vectors calculated at the first stage, using a second stage motion estimator.
FIG. 1
shows the structure of a conventional motion detecting apparatus. Referring to
FIG. 1
, a filter (FL)
501
receives and converts a video source signal into signals refi
1
and serc
1
having an input format adaptive for a motion estimator (ME)
503
. A memory (MEM)
502
is connected to the filter
501
and stores the reference signal refi
1
for I picture data and P picture data. The motion estimator
503
is connected to the filter
501
and the memory
502
and generates and outputs an address signal AD
1
shifted by a movement quantity corresponding to the motion vector MV
2
′ to the memory
502
. The motion estimator
503
calculates an absolute value of a difference between the reference signal ref
1
shifted by the movement quantity corresponding to the motion vector MV
2
′ from the start coordinates of normal retrieval and the video source signal src
1
received from the filter
501
, and accumulates the absolute values over 256 samples in units of macroblocks. The motion estimator
503
searches a motion vector (MV) when an absolute error (AE) is minimum, and adds the motion vector MV
2
′ to the searched motion vector. An addition result is then outputted from the motion estimator
503
as the motion vector MV
1
.
An memory
505
and an motion estimator
506
are connected to a delay (DLY) unit
504
and receive a video source signal delayed by the delay unit
504
. The memory
505
and the motion estimator
506
are the same in the structure as the units
502
and
503
. Similarly, the motion vector MV at a minimum of the absolute error (AE) is added with the motion vector MV
1
so that the addition result is outputted as a motion vector MV
2
.
A memory (MEM)
507
is connected to the motion estimator
506
and stores the motion vector MV
2
outputted from the motion estimator
506
. The stored motion vector MV
2
is then outputted as the motion vector MV
2
′. Accordingly, the video signal will hardly be degraded in quality while the amount of data to be calculated is significantly reduced.
However, in the conventional motion detecting apparatus, there is a possibility that a value of an erroneously detected motion vector is large when a video data changing largely and randomly over the search region of the first stage motion estimator is coded. As the coding of the large motion vector value requires a large amount of data, the ratio of coded data of the motion vector to the entire coded data will increase, resulting in degradation of the quality of the image.
In conjunction with a high efficiency vide data coding system is disclosed in Japanese Laid Open Patent Application (JP-A-Heisei 6-62392). In this reference, the high efficiency vide data coding system is composed of a 2-dimensional space filter (
10
), a subtracter (
1
) and a conversion section. The 2-dimensional space filter (
10
) controls the spatial resolution of an input video signal (a) in accordance with to the movement quantity (n) to output a video signal (b). The subtracter (
1
) outputs an interframe difference signal (c) of the video signal (b) and a video signal (m) which is obtained through a motion compensation. Also, the conversion section is composed of a quantizing unit (
3
), an orthogonal transformation unit (
2
), a motion detector (
9
) and a variable memory (
8
). The conversion section receives the interframe difference signal (c) and outputs a quantized signal (e). The motion detector (
9
) receives the input video signal (a) and a signal (t) obtained by carrying out inverse quantization transformation to the quantized signal (e) to output the movement quantity (n) to the 2-dimensional filter (
10
). The variable memory (
8
) sends out the video signal (m) to the subtracter (
1
). Thus, the spatial filter is provided in the front of the coding state.
Also, a moving object detecting apparatus is disclosed in Japanese Laid Open Patent Application (JP-A-Heisei 8-55222). In this reference, a first frame memory (
103
a
) stores a one frame of an input signal (
1
a
). A first movement detecting circuit
104
a
detects the moving object from the input signal (
1
a
) and an output signal (
1
c
) of the first frame memory (
103
). A switch (
108
) selects one image from two images based on a control signal (
1
i
) indicative of a final movement object detection result. A second frame memory stores (
1
l
) stores a signal outputted the switch (
108
). A second movement detecting circuit (
104
b
) detects the moving object from the input signals (
1
a
) and the output signal (
1
d
) outputted from second frame memory (
103
b
). A logical operation circuit (
105
) carries out a logical operation of the output signal (
1
e
) from the first movement detecting circuit (
104
a
) and the output signal (
1
f
) from the second movement detecting circuit (
104
b
). A connection processing circuit (
106
) combines moving object areas to the output signal outputted from the logical operation circuit (
105
).
Also, an image coding apparatus is disclosed in Japanese Laid Open Patent Application (JP-A-Heisei 8-102951). In this reference, an image information evaluation circuit (
50
) evaluates an amount of information of a plurality of frames of input image data stored in a frame memory (
40
). A scene change detecting circuit (
31
) detects a scene change. A DCT circuit (
14
) carries out a DCT process of the image data. A quantizing unit (
15
) quantizes DCT coefficients. A compressing method selecting circuit (
32
) selects a method of compressing based on the evaluation value and the scene changing detection output adaptively. A quantization step controller (
39
) predicts a basic quantization step of the quantizing unit (
15
) from the evaluation value and an expected amount of data obtained through the compression of one frame of the image data by the selected compressing method.
Also, a motion compensation apparatus is disclosed in Japanese Laid Open Patent Application (JP-A-Heisei 11-243551). In this reference, a motion compensation processing section (
200
) outputs a motion vector between an input image and a reference image and an estimated image obtained from the reference image in accordance with the motion vector. An absolute value difference summing section (
205
) as a distortion quantity operation section inputs the input image and the estimated image and determines a distortion quantity between the input image and the estimated image. A vector value coding section (
207
) inputs and codes the motion vector and outputs a vector code quantity. An optimal vector determining section (
210
) inputs the motion vectors, the distortion quantities and the vector code quantities, determines evaluation functions calculated from the vector code quantities and the distortion quantities to all of the motion vectors, and outputs the motion vector with a minimum value as an optimal vector.
Also, a motion detecting method is disclosed in Japanese Patent No. 2,869,142. In this reference, a temporary vector with a high similarity to a true vector is detected through block matching from sample vectors arranged in a low density spatially in a first step. A next temporary vector nearer the true vector is detected in a second step through block matching from sample vectors arranged the around
Dickstein Shapiro Morin & Oshinsky LLP.
NEC Corporation
Philippe Gims
LandOfFree
Apparatus and method for detecting motion vector in which... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for detecting motion vector in which..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for detecting motion vector in which... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3206948