Structure and method for reducing spatial noise
AssigneeMinerya System, Inc.
InventorsWilliam May, Jean-Georges Fritsch
Life of this patent
Estimating the term…
Abstract
A digital filter for noise reduction selects between local variances obtained from adjacent pixels in the same frame and adjacent pixels in the same field. In one embodiment, the digital filter includes a filter modified from an adaptive Wiener filter which preserves edges and smoothes smooth areas of the image. A high compression ratio can be achieved in very smooth regions of the image without introducing artifacts.
Claims10
A digital filter for removing noise in a video image, said video image formed by successive frames, each frame including first and second fields, each field including multiple pixels, said digital filter comprising:
- means for computing, for each pixel of a frame of said video image, (i) a first variance based on a first group of pixels, selected from said frame, adjacent said pixel;
- and
- (ii) a second variance based on a second group of pixels, adjacent said pixel and equal in number as said first group of pixels, selected from the field which includes said pixel;
- means for selecting, as a neighborhood variance, the lesser of said first variance and said second variance;
- and
- an adaptive filter, receiving said video image and said neighborhood variance, said adaptive filter having a filter coefficient which is a function of said neighborhood variance.
A method for removing noise in a video image, said video image formed by successive frames, each frame including first and second fields, each field including multiple pixels, said method comprising:
- computing, for each pixel of a frame of said video image, a first variance based on a first group of pixels, in said frame, adjacent said pixel;
- computing, for each pixel of said frame of said video image, a second variance based on a second group of pixels, selected from the field which includes said pixel and equal in number as said first group of pixels, adjacent said pixel;
- selecting the lesser of said first variance and said second variance as a neighborhood variance;
- and
- filtering said pixel using an adaptive filter having a filter coefficient which is a function of said neighborhood variance.
Dependencies are inferred from the claim text.