一种常用两次扫描算法的改进

An Improvement of Common Two-Scanning Algorithm

针对交会对接最后逼近段光学成像敏感器图像处理中的快速连通域标记问题,将标记融合和两次扫描相结合,改进了连通域标记中常用的两次扫描算法,并基于标志灯成像的几何约束和统计约束给出了可完成目标粗识别的连通域标记算法.仿真结果表明这两种改进措施都可提高有效连通域标记的效率.改进后的连通域标记算法处理一幅1 024×1 024的图像,其50次重复运行的平均耗时小于98 ms,具备实时应用的能力.

For the fast connected-component labeling problem of the optical imaging sensor for the last adjacent RVD,the two-scanning algorithm is improved via combining the procedures of labeling merge and second scanning. The fusion of target pre-recognition and the labeling algorithm is presented when considering the mutual constraint conditions and some prior knowledge. The simulation results shows that both of them can improve the efficiency of valid connected-component labeling（ CCL）. The improved CCL costs less than 98 ms in average when repetitively dealing with an image with the size of 1 024 ×1 024,and has the ability of real-time implementation.