An Improved Compressive Tracking Approach Using Multiple Random Feature Extraction Algorithm

Abstract

This paper presents an object-tracking algorithm with multiple randomly-generated features. We intent to improve the compressive tracking method whose results are fluctuated between good and bad. Because the compressive tracking method generates the image features randomly, the resulting image features varies from time to time. The object tracker might fail for missing some significant features. Therefore, the results of traditional compressive tracking are unstable. To solve the problem, the proposed approach generates multiple features randomly and chooses the best tracking results by measuring the similarity for each candidate. In this paper, we use the Bhattacharyya coefficient as the similarity measurement. The experimental results show that the proposed tracking algorithm can greatly reduce the tracking errors. The best performance improvements in terms of center location error, bounding box overlap ratio, and success rate are from 63.62 pixels to 15.45 pixels, from 31.75% to 64.48%, and from 38.51% to 82.58%, respectively.