基于标记信息的快速轮廓跟踪算法  

Quick approaches for contour following based on connected component labeling

作　　者：桑红石[1] 傅勇[2] 张天序[1] 刘云生[2] 

机构地区：[1]华中科技大学图像识别与人工智能研究所 [2]华中科技大学计算机科学与技术学院,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2005年第9期1-4,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金重大项目资助(60135020).

摘　　要：提出了三种利用标记信息加速轮廓跟踪执行过程的快速轮廓跟踪算法.算法的基本原理是:利用标记过程寻找连通区域中第一个像素出现的位置,在此基础上根据一定原则确定轮廓起点,省略了在图像中逐像素搜索轮廓起点的过程,达到加速算法执行的目的.三种算法分别适用于要求提供不同类型轮廓的应用中.仿真结果表明,大多情况下,快速算法能够将轮廓跟踪的执行时间减少40%以上.Three quick approaches for contour following based on connected component labeling were proposed. The principle of this approach is： during the second alphabetically pixel scanning of the labeling algorithm, the first pixel of each connected components is found, and a start-point table used for contour following is set up based on certain principles, thus the operation of search start points of the contour in the image is avoided, the execution of the standard contour following algorithm is accelerated. These three quick contour following algorithms are used for applications that require different kinds of contours. Simula- tion results show that the quick contour following algorithms can reduce the execution time of the standard contour following algorithm nearly 40 %.

关 键 词：图像处理 轮廓跟踪 连通域标记 快速算法 模式识别 

分 类 号：TN911.73[电子电信—通信与信息系统]