Occlusion Detection and Tracking in Image Sequence Based on the Affine Invariant
Zheng Shiyou,Fei Shumin,Yuan Xiaohui,Long Fei
DOI: https://doi.org/10.3969/j.issn.1003-0530.2006.03.034
IF: 4.729
2006-01-01
Signal Processing
Abstract:In processing the dynamic images, occlusion detection and tracking of moving targets in an image sequence are very significant and difficult problems. The objective of this research was to present a novel method to enhance the efficiency and accuracy of occlusion detection and tracking in the image sequence, and to resolve the detection difficulties brought by targets distortion while those targets moved. Firstly, the construction method of affine invariant vector was proposed, and the principles of occlusion detection based on the affine invariant vector were suggested. Secondly, the trajectory prediction technique and the occlusion area prediction method, which employed some previous information of the moving target, were presented. Finally, the detail theory and algorithm of new occlusion detection and tracking method, which were based on the affine invariant vector, were described. The results indicated that the moving objects occlusion detected by the new technique were more robust than those detected by traditional methods. In addition, the detection speed had been become more rapid. The results seem to suggest that the new method be able to provide efficiency and accuracy advantages in computer version.