论文部分内容阅读
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。
This paper presents a method of multi-maneuvering target tracking. This method uses a self-adaptive threshold to obtain a binary image, where the adaptive threshold is established by using edge and gray information, and the resulting binary image is preprocessed. On this basis, using the spatial information of the target pixel, the marker circuit is used to detect the target number in the effective field of view. At the same time, the targets are classified according to the target shape and gray level information. The target movement information is used to identify each target so as to solve the short-term target occlusion and overlap problems. The article also gives the hardware diagram and part of the software workflow for real-time implementation of this method. The method has been applied to multi-mode television tracker, and achieved the desired results.