基于色彩局域旋转搜索的区域融合方法及应用

Region Fusion of Object Based on Color Local Spiral Searching and its Application

报道了一种基于局域旋转搜索的目标区域融合方法。该方法可在搜索目标区域的同时,对那些搜索出的互不连通的各部分目标区域进行融合处理。由于融合处理只在局域范围内进行,因而处理的像素少,并可在1次局域旋转搜索过程中同时实现对目标区域的搜索及融合处理。该方法已有效地应用于彩色图像跟踪之中,实验在350MHz主频的微机上完成,系统的图像处理速度为20frames/s,相对于CCD的25frames/s图像采集速度,图像跟踪率为80%。

We present region fusion of object based on local spiral searching, in which unconnected segmented components in the object region are fused, while the search of the object region. By the local fusion, only fewer pixels have to be processed. Besides, the search and the fusion of object region are achieved in only one local spiral search. The proposed approach was applied in color image tracking effectively. Tracking experiment was achieved in a 350 MHz PC. Relatively to the CCD image collection velocity of 25 frames/s, 20 frames/s can be processed at a tracking efficiency up to 80%.