Multi-object tracking based on region corresponding and improved color-histogram matching

This paper presents a tracking algorithm based on secondary tracking strategies. First, region corresponding is used as the primary strategy. Unlike existing methods, no prediction is required in the proposed algorithm, which means high definition and low computation. Second, to deal with the corresponding problem caused by possible interaction among multiple moving objects that region corresponding approach cannot solve, further moving objects' distinguishing is carried out by the application of color-histogram matching. In order to improve the accuracy of color-histogram matching, we apply a new approach for similarity evaluation of color histograms. An intelligent surveillance system is developed to test the performance of our algorithm. Results show that the algorithm is efficient in computation and robust in various tracking tasks, and has the advantages of tracking multi-object accurately, even the shape, velocity and motion directions of objects change largely and rulelessly.