Target Tracking Algorithm Based on CUDA Parallel Acceleration
Yu Ma, Liguo Zhang, Xia Zhang · 2019
TLD is a single-target visual tracking algorithm with good robustness and high accuracy, which can solve the problems of occlusion and deformation of the target during the tracking process. However, due to the high complexity of the algorithm, it can't meet application requirements. Based on the analysis of the parallelism and complexity of the improved TLD algorithm, a parallel implementation algorithm is proposed and implemented with CUDA to improve the real-time performance of the improved TLD algorithm for the foreground detection classifier and integrated classifier in the algorithm detection module. The results show that the improved TLD algorithm based on CUDA improves the average speed of the detection module by 8.8 times and the speed is 181 frames per second. For the test videos with different resolutions, the average execution speed of TLD algorithm is 56 frames per second, which satisfies the requirement of intelligent video monitoring.