Method and system for global motion estimation and compensation

1. A method for efficient estimation and realization of inter-frame global camera motion prediction and compensation of stationary or nearly stationary objects in videos having a reference depth map and a current depth map, comprising the steps of:

detecting features in a depth map video using a modified Ramer-Douglas-Peucker algorithm;

extracting the features;

performing global motion compensation on the data from the current depth video followed by the reference depth video using template matching based on the features extracted,

estimating global motion parameters using an M-estimator Sample Consensus method;

detecting major moving objects in the depth video and estimating their parameters by comparing the current depth map and the compensated depth map from the reference image; and

estimating the motion parameters denoted as major motion parameters, and local motion compensation of macroblocks or other coding/prediction units by dividing the current depth map into non-overlapping blocks and performing the local motion correction.

2. The method of claim 1 wherein the video includes texture images and the estimation is also based on an analysis of those texture images.

3. A system for global and major objects motion estimation, compensation and depth compression, comprising:

a feature detector configured to receive at least a current depth map and a reference depth map, said feature detector acting to extract feature points from the maps and sets global motion compensation to its default values;

a global motion compensator configured to receive feature points from said feature detector when the number of feature points extracted exceeds a threshold, said compensator acting to match features from the two maps, estimate the global motion compensation and apply compensation to the features based on the estimate;

a motion detector configured to detect major moving object from the output of the compensator and estimate major motion parameters; and

a local motion compensator configured to receive global motion compensation and major motion parameters from said motion detector and output coded video information.

4. The method of claim 1 , wherein said compensated depth map is generated by wrapping said reference image by using either said estimated global motion parameters or a default global motion parameters.

5. The method of claim 1 further comprising a step of evaluating said global motion compensation by comparing a value derived from said estimated global motion parameters to a predetermined value.

6. The method of claim 5 , wherein said predetermined value is derived from said reference depth map.

7. The method of claim 5 , wherein said compensated depth map is generated by wrapping said reference image by using either said estimated global motion parameters or a default global motion parameters.

8. The method of claim 7 , wherein if said global motion compensation is not effective, said compensated depth map is generated by wrapping said reference image by using said default global motion parameters.

9. The system of claim 3 , wherein said compensation is based on either an estimated global motion parameters or a default global motion parameters.

10. The system of claim 9 , wherein the global motion compensator is further configured to adopt said compensation based on said default global motion parameters if said estimated global motion compensation is ineffective.