单晶硅超声振动辅助切削亚表面裂纹损伤检测算法

Detection algorithm of sub-surface crack damage in ultrasonic vibration assisted cutting of monocrystalline silicon

为降低加工零件亚表面损伤程度,提出基于机器视觉的单晶硅超声振动辅助切削亚表面裂纹损伤检测算法。搭建了一套单晶硅表面图像采集装置,选择合适的光源采集分辨率较高的裂纹损伤图像,使用高斯核和图像熵提取裂纹损伤特征点和特征尺度,将裂纹特征与高斯函数的二阶导结合作卷积处理,用于裂纹损伤检测,以多个特征尺度下的最大值作为检测结果。实验结果表明:设计的基于机器视觉的检测算法检测精度能够达到0.9以上,检测效果较好。

In order to reduce the sub-surface damage of machined parts,a sub-surface crack detection algorithm based on machine vision for ultrasonic vibration assisted cutting of monocrystalline silicon is proposed.A set of monocrystalline silicon surface image acquisition device was set up.A appropriate light source was selected to collect the crack damage image with high resolution.Gaussian kernel and image entropy were used to extract the crack damage feature points and feature scales,and the crack features and the second-order guided junction of Gaussian function were combined for convolution processing,which was used for crack damage detection.The maximum value under multiple feature scales was used as the detection result.The experimental results show that the detection accuracy of the designed detection algorithm based on machine vision can reach more than 0.9,and the detection effect is good.