Comparison of tracking algorithms implemented in OpenCV

Abstract

Computer vision is very progressive and modern part of computer science. From scientific point of view, theoretical aspects of computer vision algorithms prevail in many papers and publications. The underlying theory is really important, but on the other hand, the final implementation of an algorithm significantly affects its performance and robustness. For this reason, this paper tries to compare real implementation of tracking algorithms (one part of computer vision problem), which can be found in the very popular library OpenCV. Moreover, the possibilities of optimizations are discussed.