Adaptive kernelized correlation filter algorithm and application in target tracking

Abstract

Adaptive kernelized correlation Filter (AKCF) approach is designed to achieve an accurate and stable tracking for the moving target with fast motion and background clutter. The proposed algorithm combines the advantages of adaptive threshold selection method and KCF algorithm. The adaptive threshold selection method can automatically select the appropriate threshold according to the size of the object in the image. The accuracy of KCF algorithm is improved by adaptive threshold selection method. The performance of AKCF is verified by some publicly available benchmark video sequences. The experiment results demonstrate that the proposed approach which has the performance accuracy and stability can effectively realize the stable tracking for fast moving target.