Matching of video frames through coupled decomposition

In this paper an algorithm used for video matching process is discussed. This algorithm basically works on the principle of tie-point extraction. This algorithm is conceptualized to extract information from different video frames in the similar locations. This extracted information is called tie-points. Without the video matching process, most of the video processing applications cannot exist. A few applications where video matching plays a vital role is video retrieval, video registration and 3D construction. In the current methodologies, full frame matching is done, where an important parameter called spatial constraint is being neglected. The spatial constraint will give the information about the area from where the tie-points are being extracted, which is in fact going to lead the matching process. The matching of video frames through coupled decomposition is a novel technique which is introduced in this paper. This algorithm will also include the spatial constraints in the matching process along with the parameters used in the general methodologies. Though in this technique the sub frames are generated from the reference and target frames as the subsamples of conventional methods, in the generated sub frames, some specific characteristics like geometric transformation, additive noise and global radiometric differences remain consistent. They remain robust to the local changes.