基于深度学习的工件表面粗糙度等级识别方法  

A Recognition Method for Surface Roughness Grade of Workpiece Based on In-depth Learning

作　　者：陈琳[1] 朱坚民[1] 石园园 黄扬辉 Chen Lin;Zhu Jianmin;Shi Yuanyuan;Huang Yanghui(University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学,上海市200093

出　　处：《农业装备与车辆工程》2020年第10期76-79,122,共5页Agricultural Equipment & Vehicle Engineering

摘　　要：为解决变光照工件表面粗糙度等级识别困难问题,构建了粗糙度图像采集系统,设计了用于粗糙度等级识别的深度卷积神经网络。输入训练图片建立粗糙度等级模型,输入测试图片进行粗糙度等级识别。通过车削、平铣、立铣、平磨、磨外圆、研磨6种表面加工方式来验证该方法。改变光照强度采集变光照测试图片,运用颜色传递算法,以实验室标准光强下图片为源图片、变光照图片为目标图片,在不改变图片纹理情况下,实现源图片到目标图片色调颜色映射。实验结果表明,所提出方法能够解决变光照工件表面粗糙度等级识别困难问题。To solve the problem of difficult to recognize surface roughness grade of shimmering workpieces,a roughness image collection system is constructed,and a deep convolution neural network for roughness grade recognition is designed.The roughness grade model is established by inputting training picture,and roughness grade is recognized by inputting test picture.The method is verified by six surface machining methods:cutting,milling,vertical milling,flat grinding,cylindrical grinding and grinding.Change the light intensity to collect and change the light test picture,use the color transfer algorithm,take the picture under the laboratory standard light intensity as the source image to change the light image as the target image,and achieve the color mapping of the source image to the target image without changing the image texture.The experimental results show that the proposed method can solve the difficult problem of surface roughness grade recognition.

关 键 词：粗糙度测量 卷积神经网络 深度学习 图像识别 颜色传递 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]