多目标跟踪

本文提出了一种多机动目标跟踪的方法。该方法用自适应门限获得二值图像，其中，自适应门限用边缘和灰度二重信息来建立，并对由此所获得的二值图像进行预处理。在此基础上，利用目标像素点的空间信息，使用标记电路，检测有效视场内的目标个数。同时利用目标形状特征和灰度信息对目标进行分类；利用目标运动信息，对各个目标进行识别，解决短时间的目标遮挡、重叠等问题。文章也给出了实时实现该方法的硬件框图和部分软件工作流程。该方法已应用于多模式电视跟踪器，取得了预期效果。

Multiple target Tracking

This paper proposes a way of multiple maneuvering target tracking. It creates the two-value picture using adaptive threshold constructed from the information of edge and gray scale in the picture,and then,the two-value picture is preprocessed Based on this way,the number of tar-gets in the valid video-space is detected by the label-circuit to analyse spatial information of pixels on targets in the picture. At the same time,the targets are classified by utilizing the information about target shape and gray scale, and identified by using the information about target motion, so as to solve the problems while being covered or crossed in a short time. The blocks of hardware and some blocks of software are illustraled.The design has been used in multimodel TV tracker and the results are satisfactory