基于稀疏表示的目标跟踪算法分析-信号与信息处理专业论文.docxVIP

华中科技大学硕士学位论文 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract Visual tracking has been an active research topic which has a wide application in many computer vision tasks such as intelligence surveillance, vehicle navigation, human computer interaction, and so on. Sparse representation accords with peoples sense of vision characteristic and reduces the dependence on feature selection. Therefore，it has been received extensive attention in recent years. Sparse representation involves two significant components: the collection of redundant dictionary, the analysis of sparse approximation coefficients. The coefficients contain rich information including classification ，noises, which can be utilized for improving efficiency. But studies of this aspect still are very few so far. Introducing the background into the template set as the negative sample, the representative ones combine with positive samples as to build a new template set which is used to classify the sampling particles. The classification method filters out many samples which are largely dissimilar to the target, avoids the following computationally expensive cost of matching. Experiment shows that it cut down the redundant calculations. The information captured in the approximation coefficients can be utilized for occlusion analysis. From the result of experimental analysis, the coefficient distribution is in a clutter when the target is occluded by an object. It can be applied for occlusion detection. When an occlusion is detected, it prevents an improper addition to the template set and reduces the dimensions of template according to the occluded area. The experimental results indicate that the proposed method somewhat reduces the algorithm complexity and minimizes the risk of introducing the occluded target into the template set. Keywords: Visual Tracking, Sparse Representation, Particle Filter, Classification, Occlusion Detection. PAGE IV PAGE IV 目 录 摘 要 I HYPERLINK \l _bookmark0 Abstract II HYPERLIN