Image analysis – Image enhancement or restoration – Artifact removal or suppression
Patent
1994-10-13
1997-02-25
Boudreau, Leo
Image analysis
Image enhancement or restoration
Artifact removal or suppression
382261, 382274, 382107, 348607, 348701, G06K 940, H04N 521
Patent
active
056066317
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
The invention relates generally to a method for detecting and removing errors exceeding a specific contrast in digital video signals. This is achieved by using an error detector and a filter, which continuously process pixels (picture elements) associated with pictures in a picture sequence in said digital video signals.
Digital video signals usually consist of a sequence of pictures, whereat the pictures are made up of pixels of interlaced video fields. The characteristics of such digital video signals are determined by how they were originated, processed, transmitted, etc. Below, a picture will be referred to as a film frame if the original source material is film, whereas a picture will be referred to as a video field if the original source is video. Conversely, the word "picture" will in some cases mean a "field".
Dependent upon the origin, the digital video signals may contain various characteristic high contrast, peak white or peak black errors. More specifically this may refer to the presence of areas with large differences in pixel values, where the areas are adjacent within one picture, or where the areas have corresponding coordinates in different pictures.
Film originated errors, when film has been transferred to video, usually belong to two or three consecutive fields and are usually caused by scratches or dust. Such film originated errors differ from video originated errors, which are normally part of only one field, e.g. video transmission errors caused by FM-threshold noise resulting in a typical one-dimensional pattern when the picture sequence is monitored.
Known methods for detecting and removing errors exceeding a specific contrast in digital video signals include linear and non-linear processing within one picture or between pictures in the picture sequence. These methods usually include an error detection part and a filtering part. There are methods wherein the filtering part is controlled by an external triggering device.
A difficulty in known methods is that the detection is incapable of accurately separating image and error information contained in the video signals, thus giving large side-effects in loss of resolution or distortion, as well as inability to adapt the filtering in real-time according to specific shapes of errors which are to be removed.
SUMMARY OF THE INVENTION
The present invention concerns an improved method for detecting and removing errors exceeding a specific contrast in digital video signals and is specially adapted to handle a wide range of different error shapes in a picture sequence and to give a minimum of side-effects.
One particular problem to be solved by a method according to the present invention is to accurately separate signal and error information and to achieve adaptive filtering according to a current specific error shape. This problem is solved by a method characterized in that the error detector detects an error if at least one error pixel, being a pixel with a specific contrast, i.e. difference in pixel intensities, is found when comparing the pixel intensities of pixels belonging to at least two pictures in said picture sequence, and in that the error detector determines an error shape of a detected error, the error shape being a group of pixels consisting of said error pixel and adjacent pixels with a specific contrast, if any, and in that the filter is an adaptive filter which, when a current pixel is an error pixel, replaces the intensity of the current pixel with an intensity being a mathematical function of pixel intensities of pixels within and outside said error shape, whereas a current pixel that is not an error pixel is bypassed through said adaptive filter. If no pixels with a specific contrast are adjacent to the error pixel, said group of pixels includes the error pixel only.
Suitable further developments of the invention are defined in the dependent claims and will be explained below.
BRIEF DESCRIPTION OF THE DRAWINGS
A preferred embodiment of the invention will be described below with reference
REFERENCES:
patent: 4305091 (1981-12-01), Cooper
patent: 4573070 (1986-02-01), Cooper
patent: 4682230 (1987-07-01), Perlman et al.
patent: 4723166 (1988-02-01), Stratton
patent: 4825440 (1989-04-01), Heitmann et al.
patent: 4837624 (1989-06-01), Heitmann et al.
patent: 4838685 (1989-06-01), Martinez et al.
patent: 4875106 (1989-10-01), Heitman
patent: 4924306 (1990-05-01), van der Meer et al.
patent: 5099329 (1992-03-01), Oyama et al.
patent: 5276512 (1994-01-01), Onda
Christensson Bjorn
Weiss Peter
Boudreau Leo
DV Sweden AB
Mehta Bhavesh
LandOfFree
Method for detecting and removing errors exceeding a specific co does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for detecting and removing errors exceeding a specific co, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for detecting and removing errors exceeding a specific co will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1979822