A fast template matching algorithm for aerial object tracking

In this paper, a fast template matching method has been proposed based on coarse fine template matching approach. The method is aimed at aerial object tracking. In the process of coarse stage, number of sub image blocks of search area is formed adaptively based on the template size. If the object covers more than seventy percent of search region, then search area is divided into fixed size blocks. In order to choose possible blocks, threshold of zero^th order moment of template image has been compared. The suitable blocks are found out based on adjacent block numbers. During fine stage, on selected blocks, the proposed algorithms make it possible for efficient matching of template into an image based on Normalised Cross Correlation or Sequential Similarity Detection Algorithm formula. Experimental results show that proposed algorithm is very efficient and fast for image matching. The effectiveness of this method is demonstrated by object tracking of aerial targets on real videos.