基于机器视觉的轴承外圈表面图像拼接方法

Image Mosaic Method of Bearing Outer Ring Surface Based on Machine Vision

尽管基于机器视觉的缺陷检测技术已较为成熟,但应用在轴承外圈表面缺陷检测的理论研究还相对较少。应用图像拼接方法,将拍摄到的含有不完整缺陷区域的轴承外圈表面图像拼接为一幅缺陷信息完整的图像,以便后续进行缺陷的识别。在图像拼接过程中,同时采用SURF算法和图像柱面投影原理,还原了图像边缘缺陷区域的真实空间信息。建立Hessian矩阵(基于SURF算法)提取并匹配特征点,运用加权平均融合算法实现对含有缺陷信息的轴承外圈表面图像无缝拼接。经实验分析,结果表明该方法有效还原了轴承的空间几何特性,保留了边缘特征信息,能为图像拼接的准确性及拼接质量提供有效保障。无论从理论还是实践的角度来说,均有研究和探索的必要性。

Although the defect detection technology based on machine vision is relatively mature,there is relatively little theoretical research applied to the detection of defects on the surface of bearing outer rings.The image stitching method was applied to stitch the captured images of bearing outer ring surface containing incomplete defect areas into a single image with complete defect information for subsequent defect identification.In the image stitching process,both SURF algorithm and image column projection principle were used to restore the real spatial information of the defect area at the edge of the image.The Hessian matrix(based on SURF algorithm)was established to extract and match feature points,and the weighted average fusion algorithm was applied to achieve seamless stitching of the bearing outer ring surface images containing defect information.After experimental analysis,the results show that the method effectively restores the spatial geometric characteristics of the bearing and retains the edge feature information,which can provide effective guarantee for the accuracy and quality of image stitching.From both theoretical and practical perspectives,there is a need for research and exploration.