基于机器视觉的零件在线分选系统研究

Research of an online parts sorting system based on machine vision

为了完成两种工业零件在线自动分选工作,本研究利用USB相机采集传送带上零件的图像,利用大津法进行图像二值化处理,之后进行轮廓提取并完成连通区域标记操作。通过计算标记图像中两种零件的圆形度与颜色信息并与之前人工建立的标准信息进行比较,最终完成零件的分选操作。实验结果证明该分选系统可以有效降低劳动强度,提高自动化程度,分选准确率在98%以上,平均处理时间为0.51秒,能够满足实际生产中的要求。

In order to complete the work of the online sorting for two types of industrial parts, a USB camera was used to capture images of the parts on the conveyor belt. Otsu method was used for image binarization processing, then contour extraction and connected component labeling were done to the binary image. Roundness and color information of the two types of industrial parts were calculated using the labeling image, and then the sorting operation for the parts was completed by comparing the roundness and color information with the standard information which was established artificially before. Results showed that the sorting system could reduce labor intensity and improve the degree of automation; the sorting accuracy was more than 98% and it was able to meet the requirements of the actual production.