基于MATLAB图像处理的弯管检测方法  

Bend Measuring Method Based on MATLAB Image Processing

作　　者：宋洪亮[1] 王宝仁[1] 

机构地区：[1]山东科技大学机械电子工程学院,山东青岛266590

出　　处：《山东科技大学学报（自然科学版）》2017年第3期83-90,共8页Journal of Shandong University of Science and Technology(Natural Science)

摘　　要：为了提高刚性弯管角度与尺寸检测的精度与效率,弥补传统测量方法与工具的不足,提出了一种基于机器视觉的非接触式弯管检测方法。该方法通过预先设定参照工件与待测弯管,利用CCD机摄取两者含有尺寸信息的特征图像,借助MATLAB的图像处理技术提取两者的坐标数据,求解图像上参照工件的尺寸缩放倍数,进行弯管尺寸值的求解;弯管的角度与弯角半径分别采用最小二乘法线性拟合与霍夫变换拟合圆求解。实验表明:本方法检测精度优于传统检测结果;其操作简单、高效,适用于尺寸不大的各类弯管检测。To improve the accuracy and efficiency of rigid bend angle and size measurement and to make up for the weaknesses of traditional measurement methods and tools, a non-contact bend measuring method based on machine vision was proposed. By presetting the reference workpiece and bend for measurement, feature images with the size information of both the workpiece and bend was taken by CCD. By using MATLAB image processing technology, their profile coordinate data was extracted. By solving the pantograph ratio of workpiece, the bend size value was solved. The bending angle and its radius were solved by using least squares method for linear fitting and Hough transform for circle fitting. Experiments show that with simple operation, high efficiency and better accuracy than traditional methods, the proposed method is suitable for measurement of small size bends.

关 键 词：MATLAB 非接触式 图像处理 弯管检测 

分 类 号：TG703[金属学及工艺—刀具与模具]