摘要
分析了二值图像识别中常见的边界跟踪算法,在此基础上提出一种通用性强的边界跟踪算法,能够根据上一边界点的位置判断轮廓走向.在搜索下一个边界点时,只需要对候选的5个点进行判断,便可以找到下一个边界点的位置,从而减少了搜索的次数,使得边界跟踪的时间大为减少.算法对于轮廓不封闭的线段也可以一次扫描得到其轮廓信息.实验表明,算法不仅速度快,而且轮廓识别准确.对于目标物较复杂的图像,算法更能体现出其优越性.
First, the common contour detection methods in a new general boundary tracking algorithm was proposed binary image identification were analysed. Then This algorithm could judge the contour direction from the foundation of the former contour pixel. Only by judging five candidated pixels when detecting the next contour pixel, the position of the next contour pixel could be found. So the detection time could be reduced. The algorithm could also get contour information from lines which were not closed b.y scanning just for once. The experiment indicates that the algorithm is not only very fast, but the contour recognition is exact. For an image, the more the objects are complicated, the more this algorithm is advantaged.
出处
《大连海事大学学报》
CAS
CSCD
北大核心
2006年第1期62-64,67,共4页
Journal of Dalian Maritime University
关键词
图像识别
二值图像
边界跟踪
目标物体轮廓
模式识别
image identification
binary image
boundary tracking
objects＇ contour
pattern recognition
作者简介
王福生（1980-），男，辽宁大连人，研究生；E—mail：kxwfs@sina．com．
