数字全息技术的激光超精密加工表面检测研究  

Research on laser ultra precision machining surface detection based on digital holography

作　　者：周建起 刚建华 ZHOU Jianqi;GANG Jianhua(Department of Mechanical and Electrical Engineering Cangzhou normal University,Cangzhou Hebei O61001,China)

机构地区：[1]沧州师范学院机械与电气工程学院,河北沧州061001

出　　处：《激光杂志》2023年第8期243-247,共5页Laser Journal

基　　金：河北省科技厅项目(No.18211844);沧州师范学院检测技术与自动化装置科研创新团队项目(No.CXTDL1905);沧州市科学技术局重点研发计划指导项目(No.204101021)。

摘　　要：研究一种基于数字全息技术的激光超精密加工表面检测方法。使用数字全息技术生成激光超精密加工工件的立体像,将生成的立体像转化成激光超精密加工表面深度图像,利用二维Arimoto熵完成图像分割,提取出感兴趣的激光超精密加工表面缺陷目标区域,以此作为卷积神经网络输入层的输入数据,通过卷积神经网络模型构建和训练,实现激光超精密加工表面检测。实验结果表明:该方法生成的立体影像能清晰呈现激光超精密加工工件的结构和纹理特征,并且能提取完整的缺陷目标区域;在不同缺陷程度下,各工件加工表面检测效果均较为理想。In this paper,a method of laser ultra precision machining surface detection based on digital holography is studied.The 3D image of laser ultra precision machining workpiece is generated by using digital holography technolo-gy,and the generated 3D image is converted into the depth image of laser ultra precision machining surface.The image segmentation is completed by using two-dimensional Arimoto entropy,and the defect target area of laser ultra precision machining surface is extracted as the input data of the convolution neural network input layer,Laser ultra precision ma-chining surface detection is realized.The experimental results show that the 3D image generated by this method can clearly show the structure and texture features of laser ultra precision machining workpiece,and can extract the com-plete defect target area;Under different defect degree,the inspection effect of the machined surface of each workpiece is ideal.

关 键 词：数字全息 激光超精密 加工表面检测 立体像 二维Arimoto熵 卷积神经网络 

分 类 号：TN247[电子电信—物理电子学]