基于机器视觉的精密铸铁平板刮研质量检测  

Inspection of the Scraping Quality for Precise Cast Iron Plate Based on Machine Vision

作　　者：郭杰[1] 陈健生[2] 向守兵[1] 雷刚[1] 

机构地区：[1]四川工程职业技术学院,四川德阳618000 [2]清华大学电子工程系,北京100084

出　　处：《四川师范大学学报（自然科学版）》2013年第5期787-791,共5页Journal of Sichuan Normal University（Natural Science）

基　　金：国家自然科学基金(61101152)资助项目

摘　　要：针对当前机械行业检测精密铸铁平板刮研质量所存在的检测精度、准确度和效率低下的现状,研究如何运用机器视觉技术改进该检测过程.首先,获取着色对研后的平板彩色图像,经图像增强后采用K均值颜色聚类算法对图像进行分割,再按照相应标准检测各位置的接触点面积比和接触点数量,并提出以变异系数进一步量化评价分布均匀度.对颜色聚类算法进行了改进,运算速度明显提高;以10 mm为步长完成一块300 mm×300 mm的平板检测耗时约960 s,检测效率和质量远高于传统方法,可作为精密铸铁平板和精密机床导轨面的刮研质量检测的替代方法.It is necessary to study how to improve the inspection process in checking the cast iron plate scraping quality with the help of machine vision technology, in face of the fact that manual inspection usually results in lower precision, lower accuracy and lower efficiency. Firstly, take the plate image after colored scraping, and after image enhancement, segment the image by K-means color clustering algorithm; then check the proportion of bearing area and number of the high spots on each position according to the corresponding standards, and qualify the uniformity of evaluation distribution by coefficient of variation. Experiment shows by improving Kmeans color clustering algorithm the computing speed increased obviously. Time for checking a 300 mm ×300 mm plate with 10 mm step is about 960 s, the inspection efficiency and quality improved a lot than the traditional method, and the method can be widly applied for checking the scraping quality of precision cast iron plates or machine tool＇ s guideway.

关 键 词：机器视觉 图像处理 颜色聚类 铸铁平板 均匀度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]