摘要
提出了一种基于轮廓曲率及曲率累加和曲线的轮廓精确分段方法。不仅吸收了曲率类方法所具有的旋转不变性及计算量少的优点，而且吸收了直接法对单纯直线类轮廓准确分段的优点，实现对复杂轮廓的精确分段。
The segrnentation of the contour curve,or the detection of dominant points,is of greatimportance in stiape recognition,The dominant points are divided into corner points and tan-gent points, and a two-stage method based on curvature and curvature accumulation curvefor detecting the dominant points is proposed. At the first stage,the curvature is calculatedby point fitting and smoothed hy Gaussian filter and the corner points are detected by thepositive maximum and negative minimum of the curvature curve, At the second stage ,the curvature of the contour between any two consecutive corner points are accumulated and,with the commonly used splitting method,the tangent points between two consecutive cor-ner points are detected based on the curvature accumulationc:curve.
出处
《华中理工大学学报》
CSCD
北大核心
1995年第6期20-23,共4页
Journal of Huazhong University of Science and Technology
基金
国家863高技术计划项目
关键词
计算机视觉
模式识别
轮廓
曲线
分段
曲率
computer vision
pattern recognition
contpour
curve
segmentation
curva-ture:curvature accumulation
