A Panoramic Image Stabilization System Based on Block Motion Iteration

A novel approach is presented to accomplish panoramic image stabilization. Firstly, the global motion with six parameters is obtained by the iteration of block motions. In each divided image block, the gray projection correlation based on the three points´ adaptive search method is presented to obtain block motion from-coarse-to-fine. And then, all the block motions are iteratively computed with the least-square method to delete the mismatched or local motions, and so, the global motion is refined. Secondly, the Kalman filter, which can effectively alleviate unwanted camera vibrations and follow the intentional camera scan, is applied to smooth the original motion vectors. Lastly, the loss information of image boundary due to the motion compensation is reconstructed with image mosaic to prevent the visual illusion. Experimental results show that this system can smooth unwanted translation or rotation of video sequences and improve visual integrity with panoramic image compensation.