中文摘要：

复杂背景环境下，在对目标检测与跟踪过程中，极易产生光照变化、目标遮挡以及目标旋转等因素干扰。由于传统目标检测与跟踪方法，在干扰下目标检测精确度较低，导致目标跟踪性能极不稳定，甚至跟踪丢失的问题。为此提出一种基于轮廓和ASIFT特征匹配的目标检测与跟踪算法。首先，采用K—Means无监督聚类算法对视频序列图像进行分割，得到初步的分割目标；其次，在基于区域的基础上利用动态阈值OTSU法进行二值化处理，去除目标边界的噪声，经轮廓检测提取比较连续且明显的目标轮廓；最后，通过ASIFF算法提取目标轮廓特征，利用ASIFT特征匹配实现前后两帧序列图像间的精准匹配，采用CamShift算法实现目标跟踪。多组实验结果表明，改进算法提高了目标检测的效率和精度，跟踪性能稳定、可靠，实时性较好。

英文摘要：

A new algorithm about moving target detection and tracking based on contour and ASFIT feature matching is proposed. Firstly, the K - Means is used to segment the video sequence images to get the initial segmentation target. Then, on the basis of the target region, the dynamic threshold OTSU method is adopted to deal with target image binary conversion which could remove much noise at the boundary of target contour, and after that the more continuous and obvious target contour is extracted. Finally, the ASIFT algorithm is used to extract features of the target, and the accurate matching between the two frame sequences could be achieved by ASIFT feature matching, and the CamShift algorithm is used to achieve target tracking. Mutiple sets of experimental results have verified the high detection efficiency, and stable and reliable tracking performance, and strong real - time processing capability of the improved algorithm.