摘要
当前常用的连通域检测算法中很多是标号算法.为了克服标号算法的不足,提出了一种基于多线程的连通域检测算法,并考虑到图像中连通域的实际特点,构建了几个重要的连通域的数据结构,在像素之间建立一种多层链表的关系来分析和识别不同的连通区域.实验表明该算法的准确性高,计算速度比较快,计算量与标号计算法相比减小了很多.
At present, the common detecting algorithm for connected components is labeling algorithm. In order to overcome the disadvantages of labeling algorithm, this paper presents a multi-thread-based algorithm for detecting connected components, builds several key data structures for connected components considering the practical features of a connected component in an image, and analyzes and identifies different connected components by establishing a multi-layer chain structure among images. Experiment results show that this algorithm has high accuracy and fast speed of computation with less computation compared with la- beling algorithm.
出处
《重庆工学院学报》
2007年第3期80-84,共5页
Journal of Chongqing Institute of Technology
基金
重庆市教育科学技术研究项目(KJ051601)
关键词
连通区域
多线程技术
链表结构
图像处理
图像识别
connected components
multi-thread technology
chain structure
image process
image identification
