Mean-shift is a fast object tracking algorithm that only considers pixels in an object area so that it has smaller computational load. It is suitable for use in real-time conditions in terms of execution time. The use of histograms causes this algorithm to be relatively resistant to rotation and object’s size change. However, its resistance to lighting changes is not optimal. This study aims to improve the performance of the algorithm in lighting change conditions while reducing its processing time. The proposed technique includes the use of sampling techniques to reduce the number of iterations, optimization of candidate search object locations using Simulated Annealing, addition of tolerance parameter to optimize object location search and area-based-weighting instead of the Epanechnikov kernel. Using one tail t-test statistics with two independent sample groups, the test results show that the average proposed algorithm performance was significantly better than the mean-shift algorithm in terms of lighting resistance and processing time per video frame. The results of testing with 999 frames of video images give the average processing time results of the proposed algorithm's is 83.66 ms while the mean-shift algorithm is 116.86 ms.
展开▼