Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1996-08-29
2001-08-14
Britton, Howard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
Reexamination Certificate
active
06275532
ABSTRACT:
BACKGROUND OF THE INVENTION
The recent progress of high-speed digital signal processing systems with advances in LSI technology and development of video processing technique has meet an increasing interest for more effective usage of video information. In the field of telecommunications, construction and development of digital networks that is represented by ISDN (Integrated Service Digital Network) have realized video communications services such as videophones, video conferencing services, video data base services and so on. Furthermore, the spread of mobile communications networks with advances in digital information technology has increased users' need to realize mobile video communications services.
It is impractical to handle video information because the video information generally contains a very large amount of information. However, it is possible to reduce the amount of video information by reducing redundancy of the video information. High efficient video compression techniques are particularly important to the mobile communications networks that have a small amount of transmission line capacity. By this reason, international organizations ITU-T arid ISO/IEC have energetically work to settle international standards on video coding methods for encoding video signal at a very low bit-rate.
A video signal contains time-sequential information such as a change of motion in picture and spatial information concerning a content of one frame (video frame or video field; both will be hereinafter called a frame), each of them has a redundancy. A motion compensative interframe prediction with orthogonal transform encoding method that has been preferably used is such that the temporal redundancy of information is first reduced through the interframe prediction with motion compensation. According to the principle of the motion compensative interframe prediction encoding method, the motion compensative interframe predicting portion prepares predicted value of an input video signal from an already encoded signal stored in a frame memory and outputs a difference between the predicted value and the input video signal as a prediction error signal. A prediction error signal encoding portion encodes the prediction error signal by orthogonal transforming method to further reduce spatial redundancy of the information. The encoded prediction error signal is locally decoded, then stored in the frame memory and is used for prediction of a proceeding frame.
In very low bit-rate video encoding, it is necessary to represent a video-signal with a very small amount of information. An amount of information allocated to the orthogonal transform encoding (i.e., prediction error encoding), therefore, is considerably limited. Accordingly, it is very important to improve the efficiency of the interframe prediction by using an interframe prediction method that can more correctly predict a change of a video signal with time.
In the latest years, many studies have been made on interframe prediction methods using affine transformation and bilinear transformation. While the motion compensative interframe prediction method represents a movement contained in a video as a translational motion by using a motion vector per unit-area, the method using the affine transformation and bilinear transformation can more accurately represent a movement in a video, i.e., with increased prediction efficiency since it can represent movement, rotation, enlargement and deformation.
An interframe predicting portion of the conventional video-coding device comprises a frame memory portion for storing already coded video signals, a motion vector detecting portion for determining a representative motion vector per unit area from a difference between an input video signal and a video signal read from the frame memory portion, a motion vector interpolating portion for determining a motion vector per pixel from the representative motion vector and a pixel value predicting portion for preparing a predicted video signal from a video signal read from the frame memory portion by using the motion vector per pixel.
The operation of the above-mentioned interframe predicting portion is as follows:
The frame memory portion stores already coded video signals as reference video frames for interframe prediction. The motion vector detecting portion receives an input video-frame signal to be encoded and reads a reference video frame stored in the frame memory. The motion vector detecting portion divides the coding video frame into unit-areas and scans the reference video frame to find an area most similar to a current coding unit-area. A displacement of a unit-area of the coding video-frame from the area found in the reference video frame is outputted as a motion vector. The motion vector is a representative motion vector representing a interframe displacement of a representative point within a unit-area (usually, a center of the unit-area). The relationship between a representative point and a vector searched unit-area are specified. In searching a similar area in a reference video frame, a sum of differential absolute values or a square-sum of differential values of each pixel in a unit-area is used as a scale of similarity. Furthermore, a center portion of the unit-area may be weighted by multiplying a differential pixel value at the center portion of the unit-area by a large coefficient and a differential pixel value at the periphery thereof by a small coefficient prior to summation so as to more accurately determine a displacement of the representative point.
The representative motion vector is transferred to the motion vector interpolating portion which in turn determines motion vectors for respective pixels using the received representative motion vectors. For affine transformation, a motion vector for each pixel existing within a triangle area surrounded by three neighboring representative points (hereinafter called transformable unit-area) is determined by solving an affine transformation expression from the representative motion vectors of the respective representative points. For bilinear transformation, a motion vector for each pixel existing within a quadrangular area surrounded by four neighboring representative points (hereinafter called transformable unit-area) is determined by solving a bilinear transformation expression from the representative motion vectors of the respective representative points. For a transformation unit-area being quadrangular or rectangular, it is equivalent that a motion vector value of each representative point is distributed in a vertical direction and a horizontal direction in proportion to an a distance between a remarked pixel and a representative point.
The pixel value predicting portion receives the (remarked) pixel motion vectors inputted pixel by pixel and, considering the motion vectors to be a displacement of the remarked pixel from its corresponding position in the reference frame, reads a pixel value at the corresponding position from the frame memory as a predicted value of the remarked pixel to compose a predicted frame. If the pixel value indicates a position in the reference frame, where no pixel exists, e.g., the pixel motion vector (displacement) is a decimal fraction, a neighboring pixel value in the reference frame is read-out from the frame memory and a predicted value of the remarked pixel is determined as an interpolating value according to a known bilinear interpolation method.
In a simplified structure of an interframe predicting portion of a conventional video-decoding device using affine and bilinear transformation method, the interframe predicting portion of the conventional decoding device comprises a frame memory portion for storing already decoded video signals, a motion-vector interpolating portion for determining a motion vector of each pixel from a representative motion vector inputted for each unit-area, a pixel value predicting portion for preparing a predicted video signal from a video signal read from a frame memory by using the pixel motion vectors.
The conventi
Ema Nobuyuki
Hibi Keiichi
Britton Howard
Sharp Kabushiki Kaisha
LandOfFree
Video coding device and video decoding device with a motion... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Video coding device and video decoding device with a motion..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Video coding device and video decoding device with a motion... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2493318