Method of estimation and of hierarchised coding of the motion of

Details Method of estimation and of hierarchised coding of the motion of Method of estimation and of hierarchised coding of the motion of

1993-07-07

2000-08-01

Chang, Jon

Image analysis

Applications

Motion or velocity measuring

382236, 382240, 348416, G06T 720, G06T 900

Patent

active

060978325

DESCRIPTION:

BRIEF SUMMARY
BACKGROUND OF THE INVENTION

1. Field of the Invention
The present invention relates to a method of estimation and of hierarchised coding of the motion of image sequences, used in particular in image transmission apparatuses with a view to decreasing the information rate.
2. Discussion of the Background
A method and a device for estimation and hierarchised coding of the motion of image sequences have been proposed in French Patent Application No. 89 11328 filed in the name of THOMSON CONSUMER ELECTRONICS, which corresponds to U.S. patent application Ser. No. 07/678,278, now allowed. This method is characterised in that it consists, at a first level, in partitioning the current image into macroblocks of 2.sup.p+1 .times.2.sup.p+1 pixels and in determining a first motion-vector field associated with this division into macroblocks, by using an iterative and recursive estimation algorithm initialised with motion vectors estimated for the previous image, then, at a second level, in partitioning each of the macroblocks into quadrants and in determining, for the blocks which result therefrom, a second motion-vector field by using the same estimation algorithm but initialised with vectors from the field of motion which were estimated at the previous level, then at i.sup.th level, i=3 to p, in partitioning each of the blocks considered at level i-1 into quadrants and in determining, for the blocks which result therefrom, an i.sup.th motion-vector field by using the same estimation algorithm initialised with the vectors from the field of motion which were estimated at the previous level, the blocks of minimum size being blocks of 2.sup.1+1 .times.2.sup.1+1 pixels, and in determining a final motion-vector field from the p resulting motion-vector fields by choosing the lowest level of division for which the motion vector associated with the corresponding block leads to the minimisation of a criterion expressing the differences in luminance between blocks corresponding in the successive images through the estimated displacement vectors.
This process of estimation makes it possible to enhance the convergence of the recursive motion-estimation algorithm used and makes it possible optimally to adapt the motion-estimation algorithm to the "Quadtree" type coding which will be done subsequently of the resulting motion field.
This process of estimation can be used, among other things, in the context of HD-MAC.


SUMMARY OF THE INVENTION

The purpose of the present invention is to afford various improvements to the method of estimation and of hierarchised coding of the motion of image sequences, described in French Patent Application No. 89 11328.
Accordingly, the subject of the present invention is a method of estimation and of hierarchised coding of the motion of image sequences, consisting: 2.sup.p .times.2.sup.p pixels and in determining a first motion-vector field associated with this division into macroblocks, by using an iterative and recursive estimation algorithm initialised with motion vectors estimated for the previous image, quadrants and in determining, for the blocks which result therefrom, a second motion-vector field by using the same estimation algorithm but initialised with vectors from the field of motion which were estimated at the previous level, considered at level i-1 into quadrants and in determining, for the blocks which result therefrom, an i.sup.th motion-vector field by using the same estimation algorithm initialised with the vectors from the field of motion which were estimated at the previous level, the blocks of minimum size being blocks of 2.sup.p .times.2.sup.p pixels, motion-vector fields, characterised in that the motion-vector field is determined by using the differences in luminance between blocks corresponding in the successive images through the estimated displacement vectors of the p levels of division corresponding to each block of minimum size and by choosing as motion vector associated with the block of minimum size the one for which the difference in luminance is a minimum over this

REFERENCES:
patent: 4849810 (1989-07-01), Ericsson
patent: 4853775 (1989-08-01), Rouvrais et al.
patent: 5089887 (1992-02-01), Robert et al.
patent: 5144423 (1992-09-01), Knauer et al.
IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 11, No. 7, Jul. 1989, New York, US, pp. 742-748, P.Strobach "Quadtree-Structured Linear Prediction Models for Image Sequence Processing.".

Affiliated with

Also associated with

Rating

Method of estimation and of hierarchised coding of the motion of 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 estimation and of hierarchised coding of the motion of, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of estimation and of hierarchised coding of the motion of will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-671645