摘要
以在线生产的热钢坯标号为识别对象,设计了一套实时视觉检测系统。利用灰度面积滤波、改进Marr边缘提取算子、自动阈值等图像处理技术滤除干扰,完成字符提取;通过投影法得到单个字符;并且引入距离变换的思想对模板匹配法进行改进,能更加准确地识别标号。视觉系统还利用无线技术进行数据传输。实验结果表明该检测系统是有效可行的,在识别准确率上近98%,每块钢坯的识别时间约1.5s,能满足生产线需要。
In order to recognize characters on steel billets, a real-time automatic VisiOn inspection system was designed. Based on image processing technique including grayscale area filter, improved Marr edge detector and auto-threshold algorithm, noises were filtered and characters were extracted from the image. Single character was achieved by using projection function. An improved template match technique on distance-transforrn was used to recognize the characters. Wireless technique was applied to transfer information. The experimental result shows that this system is feasible and reasonable, the recognition rate is near 98%, and the recognition time spent on one billet is less than 1.5 second.
出处
《传感技术学报》
EI
CAS
CSCD
北大核心
2006年第3期686-689,共4页
Chinese Journal of Sensors and Actuators
基金
浙江省重点科研资助项目(2006C21037)
关键词
自动视觉检测
模板匹配
图像分割
字符识别
automatic vision inspection
template match
image segmentation
character identification
