摘要
为实时监测折页加工过程,用CMOS摄像头组装了一个简易的折页机视觉系统,进行折页错位误差的检测。为最大限度地利用现有摄像头的性能,提高检测的精度和速度,仅检测折页后幅面的一部分。同时,对软件检测折页错位,采用了基于特征点和直线相交两种方法,进行了折页错位检测实验。得到在计算机主频2.8GHz、内存2MB,摄像头分辨率640pixel×480pixel、帧率50f/s的条件下,两种方法的处理时间分别约为150ms和430ms。从实验效果看,该系统可分辨0.5mm的折页错位误差,表明机器视觉可用于折页错位实时检验,基于直线相交的方法抗干扰性好,但处理时间稍长,有进一步改进的余地。
For real-time monitoring of folding processes, the camera of CMOS is used to assemble a simple folding vision system which detect the error of folding dislocation. To utilize the camera performance completely, improve the accuracy and speed of detection, only a part of a sheet after folding will be tested. Meanwhile, the software detects folding dislocation, using methods of feature-based point and the straight line inter- secting. While the computer clock speed is 2. SGHz, the memory is 2MB, camera resolution is 640pixelx480pixel, and frame rate is 50f/s, the processing time of two methods is ap- proximately 150ms and 430ms. The result shows that folding dislocation can be distinguished from the error of 0.5mm. It can be concluded that machine vision can be used to inspect for real-time in folding. Because the interference of a straight line intersecting is good, and the processing time is slightly longer, it can be further improved.
出处
《北京印刷学院学报》
2011年第4期33-36,共4页
Journal of Beijing Institute of Graphic Communication
基金
北京印刷学院引进人才科技研究项目资助(Ed-07-17)
关键词
折页
机器视觉
图像检测
折页错位
folding
machine vision
image processing
folding dislocation
