Television – Image signal processing circuitry specific to television – Motion vector generation
Patent
1992-11-19
1995-02-28
Britton, Howard W.
Television
Image signal processing circuitry specific to television
Motion vector generation
348416, H04N 7137
Patent
active
053941969
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to the processing of images, in particular to the processing of video images such as television type images.
2. Discussion of Background
In general, in the case of video images, in particular in the case of television images, most commonly sequences of images are displayed representing scenes in motion. Therefore, the pixel should no longer be regarded as a simple image element, but rather as an object element, this object element being able to evolve within the image over time, that is to say appear, move, change luminance and/or disappear. It is therefore essential to be able to take this motion into account at pixel level, this motion being a very frequent cause of change over time. Taking the motion into account therefore makes it possible efficiently to process most of the possible situations within a scene.
To carry out a scene analysis as correctly as possibly, it is therefore important to be able to classify the pixels by assigning them a tag giving information about the evolution of the motion of the pixel. The tags giving a correct scene analysis are the tags chosen from those belonging to one of the classes termed "normal", "occluding", "appearing" or "disappearing". Indeed, the "normal" tag signifies that the image element is present in the two successive images considered, the "appearing" tag signifies that the image element is present in the current image but absent from the preceding image. The "disappearing" tag signifies that the image element is absent from the current image but present in the preceding image, whereas the "occluding" tag signifies that the current image contains foreground objects which occlude background zones of a preceding image. These four classes correspond perfectly to the various events which may occur at pixel level.
The assigning of a tag to each pixel of an image therefore shows itself as an indispensable complement to the estimation of motion and to the interpolation of images used in the context of video image processing, in particular to the temporal interpolation of images, as described in French Patent Application No. 87 07814 in the name of THOMSON GRAND PUBLIC or to the temporal interpolation of images with compensation for corrected motion, as described in French Patent Application No. 89 04256 filed in the name of Thomson Consumer Electronics.
Indeed, although interpolation with compensation for motion makes it possible accurately to reconstruct the pixels in motion, problems remain at the level of the pixels appearing or disappearing from one image to the next. In general, the estimator is perturbed by the presence of such pixels which have no counterpart in the other image. Moreover, the interpolation of the pixels should only be performed on the basis of the input image which contains them, thus constituting a particular procedure which can only be applied if the pixels have been identified.
SUMMARY OF THE INVENTION
Accordingly, the aim of the present invention is to rectify these disadvantages by providing, firstly, a method of classifying the pixels of a current image which make it possible to detect the pixels or image elements appearing and disappearing from one image to the next in the case where the current image is obtained by temporal projection of the field of motion of the preceding image belonging to a sequence of moving images.
Thus, the subject of the present invention is a method for classifying the pixels of a current image t(j+1) obtained by temporal projection of the field of motion of a preceding image t(j) belonging to a sequence of moving images, the pixels of the preceding image possessing a motion vector and a tag, the tag belonging to a class defined from at least three classes termed "normal", "occluding" or "appearing", the said method being characterised in that:
before projection, all the pixels of the current image t(j+1) are tagged "appearing";
then, during temporal projection carried out in a known manner, the tagging is modified by t
REFERENCES:
patent: 5214751 (1993-05-01), Robert
IEEE Transactions on Communications, vol. 37, No. 12 Dec. 1989, (New York, US), S. C. Brofferio: "An Object-background image model for predictive video coding", pp. 1391-1394.
BBC Research Department Report, No. 11, Sep. 1987, (Tadworth, GB), G. A. Thomas: "Television motion measurement for DATV and other applications", pp. 1-20.
Signal Processing, vol. 15, No. 3, Oct. 1988, (Amsterdam, NL), M. Hotter, et al.: "Image segmentation based on object oriented mapping parameter estimation", pp. 315-334.
Frequenz, vol. 43, No. 5, May 1989, (Berlin, DE), M. Gilge et al.: "Codierung von farbigen Bewegtbildszenen mit 64 kbit/s-Ein neuer Ansatz zur Verwirklichung eines Bildtelefons im ISDN (Teil III)", pp. 126-133.
"Thomson-CSF"
Britton Howard W.
LandOfFree
Method of classifying the pixels of an image belonging to a sequ 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 of classifying the pixels of an image belonging to a sequ, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of classifying the pixels of an image belonging to a sequ will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-851624