摘要：
为对汽车卡钳排气螺钉的微小螺纹尺寸实现高精度在线检测,提出一种基于改进SUSAN算法的卡钳排气螺钉参数辨识方法.首先,对经过兴趣域提取的螺纹图像进行二值化及边缘保持滤波处理,减小光线、噪声等对图像的干扰;SUSAN算法是采用一个近似圆形的模板在图像上移动,寻找出模板内部每个图像像素点的灰度值与模板中心像素的灰度值相同或相似的区域,再根据区域大小判断出角点位置,运用Forstner算子可进一步获得准确的角点坐标,从而计算出M10螺纹大径、中径、小径、螺距及牙型角等几何参数;利用该算法设计一套基于机器视觉螺纹检测系统,并利用万工显与该方法进行比对实验.实验结果表明:该方法的螺距、大径、中径、小径的测量精度为0.01 mm,牙型角精度为8',均满足螺纹紧固件的测量精度要求,且比传统算法精度高.%In order to achieve high-accuracy online detection of small-sized thread dimension of caliper exhaust screw, an identification method of caliper exhaust screw parameters based on improved SUSAN algorithm was proposed. Firstly, reduced the light, noise and other interferences on image after the binaryzation and edge filtering processing for image extracted from region of interest; SUSAN algorithm was an approximate circular template moving on the image to find out the region with pixels which had same or similar gray values of each pixel in the template with the template's center pixel, and then the precise angular coordinate could be calculated by using the Forstner operator so as to figure out the accurate major diameter, pitch diameter, minor diameter, pitch and tooth angle and geometric parameters of M10 thread; Moreover, the algorithm could be used to deisgn a set of detection system based on mechanical vision thread and universal tool microscope could be used to comparision test for the method. The results indicate that the measurement accuracy for major diameter, pitch diameter and minor diameter of screw pitch is 0.01mm, including 8' for thread angle, which is in conformity with requirements for measuring accuracy of threaded fastener, and has a higher precision than traditional algorithm.
