| Target tracking was a very important research topic in the field of computer vision.Although the research of target tracking had made great breakthroughs,there were still many challenging problems to be solved.This paper focused on the application of visual attention mechanism in video target tracking.The results were as follows:1)As a long-time single target tracking algorithm,TLD algorithm was a long time single target tracking algorithm,but the tracking effect was not ideal when the target was occluded,fast rotation and other problems.Aiming at this problem,the TLD algorithm was improved by taking video target tracking as the research object.In the tracking module of TLD algorithm,Shi-tomas algorithm was used to obtain feature points instead of uniformly selected points in optical flow method.The improved algorithm had good stability.2)Aiming at the problem of occlusion and morphological change in TLD target tracking algorithm,a TLD tracking algorithm based on visual attention mechanism was proposed.Through extracting visual salient regions,the color histogram of the extracted salient regions was calculated,the target model and candidate model were constructed,and the number of high-quality samples was increased.The simulation results showed that the improved algorithm the tracking accuracy and stability.3)Aiming at the problem of low real-time performance and accuracy in sparse representation target tracking algorithm,an object tracking algorithm based on sparse representation of visual attention mechanism was proposed.Firstly,multi-modal fusion of candidate samples was performed,and then visual attention mechanism was integrated into sparse representation algorithm.This paper mainly focused on the tracking problem of target in complex environment.The traditional tracking algorithm incorporates visual attention mechanism,improved the original algorithm,the stability and accuracy of tracking,and provided a new idea for target tracking algorithm.Figure 29;Table 2;Reference 51... |
PDF Full Text Request