摘要
本文针对含圆形物体的一类数字图象,提出一种采用圆弧法线参数回归方法的快速处理识别算法。该方法对圆形图案和物体等图象进行处理,识别圆心位置和半径尺寸,获得满意的实验效果,并在新日铁的机器人视觉系统中用于识别带钢卷。
This paper describes a method of recognizing circular objects in gray images. First, we detect ridge line after enhancement of edges in original images by dif- ferential filter. Secondly, we segment tracked curves into simple arcs by means of estimating tangents and normals and analysing curvature function of the curves, 'Normal regression method' is adopted in order to find optimal center point of each arcs. Finally, the true center points and radiuses are detected by clustering all arcs according to priority, center points and radiuses, we used this method to process gray images which include circular figures and strip steel coils, and have achieved satisfactory results in the experiment
关键词
数字图象
圆形物体
识别
图象处理
image processing
recognition
circles pattern
regression analysis/borders detection
