基于机器视觉的齿轮产品外观缺陷检测

提出应用机器视觉技术对齿轮外观缺陷进行检测.首先对齿轮产品实物图像进行预处理,提取齿轮边缘图像,对其实物图像和设计图像进行分割和缩放处理,得到相同大小的实物;其次,采用互信息方法对两幅图像进行配准,通过不断旋转实物图像,计算它与设计图像的互信息,找出实物图像的最佳旋转配准角度;在此基础上将两幅图像进行小波融合,即自动对齐叠加在一起,根据重合效果,检测人员可以判断生产出的齿轮产品是否合格.通过实例验证了该方法的可行性和有效性,检测效果直观.

Defect Inspection of Gear Product Appearance Based on Machine Vision

The application of machine vision technology was proposed to inspect the gear product appearance defect.First,the edge image of gear was extracted from the preprocessed image of gear product.Secondly,segmenting and scaling operation were done to the product image and designed image of gear.Thirdly,the two images were registered by the method of mutual information.Through rotating the image of gear product,the similarity of the two images was computed by the measurement of mutual information,and the best matched position could be found where they had the most mutual information.Finally,the two images were fused into a fusion image based on wavelet transform.The inspector could test the qualification of gear product by the fusion image.The feasibility and effectiveness are proved by the examples and the inspection results can be easily observed.