基于多特征融合的camshift跟踪算法研究-光学工程专业论文.docxVIP

algorithm which is based on feature fusion and motion estimation. CamShift algorithm only uses the color feature for the target modeling, does not use the other features containing the target information. In view of this, the paper proposed a CamShift tracking algorithm based on feature fusion and motion estimation. Weightedly fuse the color information and the edge information to form the probability distribution of the target model, using second-order autoregressive model to correct target position and velocity. Then in the LSV combination color space, take the SURF method to correct the region of interest to improve the accuracy and stability of tracking.Finally, using VC + + and OpenCV complete realization of the tracking algorithm, and design experiments and analysis for the algorithm. The experimental results show that, compared with the classic CamShift algorithm and similar improved algorithms, the proposed algorithm can achieve real-time tracking of the rotation and zoom moving target against complex backgrounds, and has excellent anti-interference ability.KEY WORDS：target tracking, CamShift algorithm, SURF, feature fusion目录目录 .......................................................................................................................................... 4第一章 绪论 .................................................................................................................................... 11.1 课题背景及研究意义 ....................................................................................................... 11.2 国内外研究现状 ............................................................................................................... 21.3 目标跟踪算法概述 ........................................................................................................... 31.4 本文的研究内容与结构 ................................................................................................... 5第二章 基于均值漂移的视频目标跟踪..................................................................