二值图像连通域的分段标记算法及实现

Segment labeling algorithm of connecting area in binary images and its realization

在基于图像的精确制导中对标记算法的时间性能要求很高,常规的标记算法无法满足高速实时图像处理的需要。在分析现有算法的基础上,引入段的概念,以段为标记单位记录连通区域,在对图像一次扫描的过程中完成连通性传递闭包的计算。在标记的同时记录标记目标的边沿像素坐标值,计算目标的形心、面积、轨迹、灰度等特征。给出了算法的思路和实现步骤。使用该算法标记128×128的二值图像,用TMS320C6414 600 MHz simulator,可以在0.4 ms内完成。将文中算法与其他4种连通区域标记算法进行时间性能对比,明显优于其他算法。软件仿真结果也显示分段标记算法的正确性与稳定性。适用于高速运动目标的实时截获与跟踪。

The time performance of region labeling algorithm is very important for precision guidance based on image.However,common region labeling algorithms can not meet the requirements of real-time image processing.By analyzing the current image labeling technologies,a technique using segment to record the connective area was proposed in this paper,connective closure and information related to the target could be computed during a one-time image scan.The coordinates of target edge pixels were recorded to computer target＇s features such as centroid,area,track and grey etc.as labeling.Then,the algorithm＇s conceptions and realization step were presented.Experimental results show that the proposed algorithm is capable of labeling targets in the image of 128 ×128 size in 0.4 ms with TMS320C6414 600 MHz simulator,and is more efficient,correct and stable compared with the other algorithm.It can well meet the timer performance requirements of real-time processing,and is applicable in high speed target capture and tracking.