融合单应性约束SIFT特征匹配的轴承滚子检测

Bearing roller detection based on SIFT feature matching with homography constraints

针对实际工业生产中轴承滚子自动检测时可能存在的误检问题,提出了一种融合单应性约束尺度不变特征变换(SIFT)的轴承滚子检测方法。首先,利用二维伽玛矫正对轴承原始图像进行图像增强处理,以改善后续阈值分割的效果,同时运用局部自适应分割算法对增强后的图像进行阈值分割并对分割结果进行形态学处理;在此基础上,使用边界跟踪算法获取轴承二值图像的轮廓,并截取轮廓在原始图像中对应的区域,便于利用FANN算法对该区域的SIFT特征与轴承滚子模版SIFT特征进行匹配,再进一步融合单应性约束策略对匹配点对进行筛选,以得到轴承滚子检测结果;最后,通过实验发现,所提融合单应性约束SIFT特征匹配方法的正确检测率比传统阈值分割法提高了33%,从而验证了该方法的有效性与可行性。

Aiming at the problem of misdetection in automatic detection of bearing rollers in actual industrial production,a bearing rollers detection method based on SIFT feature matching with homography constraints is proposed. Firstly,the original image of bearing is enhanced by two-dimensional gamma correction to improve the effect of subsequent threshold segmentation. At the same time,the local adaptive segmentation algorithm is used to segment the enhanced image by threshold,and the segmentation results are processed by morphology. On this basis,boundary tracking algorithm is used to obtain the outline of the binary image of bearing. The region corresponding to the contour in the original image is intercepted to match the SIFT feature of the region with the SIFT feature of the bearing roller template by FANN algorithm,and then the matching point pair is filtered by the homography constraint strategy to get the bearing roller detection result. Finally,the experiment shows that the fusion proposed in this paper is feasible. The correct detection rate of homography constrained SIFT feature matching method is 33% higher than that of traditional threshold segmentation method,which verifies the effectiveness and feasibility of the method proposed in this paper.