Image analysis – Image segmentation
Reexamination Certificate
1996-12-31
2002-04-30
Mehta, Bhavesh (Department: 2621)
Image analysis
Image segmentation
C382S232000, C382S233000
Reexamination Certificate
active
06381364
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the field of picture/video encoding/decoding. More specifically, the present invention relates the art of content sensitive picture/video encoding/decoding.
2. Background Information
A number of picture/video encoding/decoding techniques are known in the art, for examples, the encoding/decoding approaches in accordance with the standards jointly promulgated by the International Standard Organization (ISO) and International Electrotechnical Committee (IEC), formally known as ISO/IEC 11 172 and 13 818, commonly known as MPEG-1 and MPEG-2. In each of their attempts to facilitate subsequent faithful reproduction of the original picture, these prior art approaches typically employ various transformations to eliminate what's considered to be redundant or insignificant picture data. A mean square error type technique is typically employed to minimize the cumulative pixel differences between the reproduced and the original pictures, without any regard to the contents of the picture.
Experience has shown that such pixel difference minimization approach works well for objects with structured texture, such as man-made objects like a table, a chair and so forth. However, for objects with random texture, such as natural objects like a body of water, a field of flowers etc., their visual quality appear to suffer. The reason is because the human eye favors the randomness characteristics in objects with random texture. Thus, pictures that preserve the randomness characteristics for these objects are perceived by the human eyes to be of higher quality. Unfortunately, the prior art imperfect approach of trying to faithfully preserve and reproduce the pixels tend to destroy the randomness characteristics of these objects. Thus, an improved picture/video encoding/decoding technique that takes into account the content texture of the pictures/video frames is desirable.
SUMMARY OF THE INVENTION
A picture/video encoder is provided with operating logic for segmenting objects with random texture from objects with structured texture in a picture/video frame, and generating random function parameters for the objects with random texture. The random function parameters are used subsequently to regenerate the objects with random texture, when reproducing the picture/video frame. In one embodiment, the two types of objects are separated by recursively decomposing each picture/video frame into sub-blocks, separating the random sub-blocks from the non-random sub-blocks. In one embodiment, randomness is measured in accordance with a number of statistical moments determined for each of the sub-blocks. In one embodiment, the statistical moments are output as the random function parameters for subsequent regeneration of the sub-blocks. In another embodiment, noise functions are employed to recursively determine the random function parameters for the random sub-blocks.
Additionally, a picture/video decoder is provided with operating logic for regenerating objects with random texture in accordance with the random function parameters provided, and for reassembling the regenerated objects with random texture with decoded objects with structured texture to reconstitute the original picture/video frame. In one embodiment, the objects with random texture are regenerated by regenerating random sub-blocks, and the original picture/video frame is reconstituted by combining the random sub-blocks with the non-random sub-blocks of the decoded objects with structured texture. In one embodiment, the random sub-blocks are re-determined using random signals and probability density functions in accordance with the provided statistical moments, whereas in another embodiment, the random sub-blocks are regenerated using noise functions in accordance with the function parameters provided.
REFERENCES:
patent: 4941193 (1990-07-01), Barnsley et al.
patent: 5214507 (1993-05-01), Aravind et al.
patent: 5394191 (1995-02-01), Sirat et al.
patent: 5561464 (1996-10-01), Park
patent: 5815670 (1998-09-01), Iverson et al.
patent: 5915036 (1999-06-01), Grunkin et al.
patent: 5995651 (1999-11-01), Gelenbe et al.
patent: 6201879 (2001-03-01), Bender et al.
“Pyramid-Based Texture Analysis/Synthesis”, David J. Heeger, James R. Bergen, SIGGRAP 95, Conference Proceedings, Aug. 6-11, 1995, pp. 229-238.
“Pyramid-Based Texture Analysis/Synthesis”, David J. Heeger, James R. Bergen, SIGGRAPH 95, Conference Proceedings, Aug. 6-11, 1995, pp. 229-238.
Bayat Ali
Blakely , Sokoloff, Taylor & Zafman LLP
Intel Corporation
Mehta Bhavesh
LandOfFree
Content texture sensitive picture/video encoder/decoder does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Content texture sensitive picture/video encoder/decoder, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Content texture sensitive picture/video encoder/decoder will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2884079