US 5,852,475 AGranted

Transform artifact reduction process

AssigneeCompression Labs, Inc.

InventorsSmita Gupta, Yi Tong Tse

Life of this patent

Estimating the term…

Loading patent…

Abstract

A post-processor for a decoded video sequence includes a digital noise reduction unit and an artifact reduction unit which significantly reduce blocking artifacts and mosquito noise in a video image. The post-processor uses both temporal and edge characteristics of the video image to enhance the displayed image. The post-processor operates on a current frame of pixel data using information from the immediately preceding post-processed frame that is stored in a frame memory of the post-processor. The post-processor first identifies texture and fine detail areas in the image. The post-processor uses artifact reduction only on portions of the image that are not part of an edge, and are not part of a texture or fine detail area. Since artifact reduction is not utilized on these areas, the post-processed image is not softened in regions where it is easily noticed by the human eye.

Claims111

Dependent claims

A transform artifact reduction method for decoded video pixel data, said method comprising:

digitally noise reducing a block of pixels to obtain a digitally noise reduced block of pixels;
and
filtering said digitally noise reduced block of pixels using a spatially-variant filter wherein said filtering further comprises
adaptively identifying each pixel in said digitally noise reduced block of pixels as one of an edge pixel and a non-edge pixel;
assigning a pixel in said digitally noise reduced block of pixels one classification in a plurality of edge classifications using said edge pixel and said non-edge pixel identifications for pixels in a window about said pixel, said assigning said pixel in said digitally noise reduced block of pixels one classification in said plurality of edge classifications using said edge pixel and said non-edge pixel classifications for pixels in a window about said pixel further comprising
generating an edge map for said digitally noise reduced block of pixels wherein in said edge map, an edge flag for each pixel in said digitally noise reduced block of pixels is configured to identify each pixel as one of said edge pixel and said non-edge pixel;
and
adaptively filtering each pixel in said digitally noise reduced block of pixels based upon said assigned classification to reduce transform artifacts in a video image.

5 more elements

Dependencies are inferred from the claim text.