基于预测匹配差与全局-局部阈值化的轴承缺陷检测与定位算法

发布时间：2018-08-26 07:43

【摘要】：为了准确检测轴承在生产加工过程中出现的滚动体漏装等缺陷,提出了基于预测匹配差与全局-局部阈值化的轴承缺陷检测与定位算法,完成滚动体缺失、破损检测与定位。首先,引入分段线性图像增强技术,扩大滚动体与轴承背景的对比度;其次,综合全局与局部阈值化方法,结合种子填充技术,对轴承进行连通边缘标记;再设计一种圆验证机制,将轴承中的非圆边缘滤除,以提取滚动体的ROI区域,缩小了目标检测范围,提高滚动体缺陷的检测效率;最后,利用Open CV来统计不同部件的轮廓面积,从而设计预测匹配差方法,对缺失或破损的滚动体进行定位。仿真结果显示,与当前轴承检测方法相比,对于滚动体漏装或破损轴承,所提算法具有更高的检测与定位准确。
[Abstract]:In order to accurately detect the defects such as missing rolling body in the process of production and processing, a bearing defect detection and location algorithm based on prediction matching difference and global-local threshold is proposed to complete the rolling body missing, damage detection and location. Firstly, a piecewise linear image enhancement technique is introduced to enlarge the contrast between the rolling body and the bearing background. Secondly, the connected edge of the bearing is marked by combining the global and local thresholding method and seed filling technique. Then a circular verification mechanism is designed to filter the non-circular edge of the bearing to extract the ROI area of the rolling body, which reduces the detection range of the target and improves the detection efficiency of the rolling body defect. Finally, the contour area of different parts is counted by using Open CV. Thus, the method of prediction matching difference is designed to locate the missing or damaged rolling body. The simulation results show that the proposed algorithm is more accurate than the current bearing detection method for the missing or damaged rolling bearing.
【作者单位】：西安理工大学工程训练中心;
【基金】：国家自然科学基金项目(61405157,61075007) 陕西省自然科学基础研究计划(2012JM7006)
【分类号】：TH133.3;TP391.41

【相似文献】