多特征实时电视跟踪器

Real-Time Multi-Features TV Tracker

本文描述一种多特征实时电视跟踪器,应用3×3中值滤波器对目标图像进行预处理,采用SOBEL算法提取目标的边缘信息,利用目标灰度和边缘联合特征的贝叶斯分类算法检测和分割目标,在低对比度小目标情况下可以稳定跟踪目标,经外场试验取得满意的结果,该电视跟踪器已用于地空武器系统中。

A real-time multi-features TV tracker is described. The target image is preprocessed by using a 3 x 3 median filter algorithm, the target edge information is extracted by using the SOBEL algorithm. According to the target grey-level and edge association features, the target can be detected and segmented by Bayesian classifier algorithm. The TV tracker can steadily track small targets of low contrast. Satisfactory results have been achieved in battlefield experiment.