CH376纠正机器视觉提取螺孔误差的方法  

Method of CH376 in correcting extraction error of screw hole by machine vision

作　　者：石小仕 赵良军 姚娅川 王雷 董林鹭 SHI Xiaoshi;ZHAO Liangjun;YAO Yachuan;WANG Lei;DONG Linlu(College of Automation and Information Engineering, Sichuan University of Science & Engineering, Zigong, Sichuan 644004, China;Computer College, Sichuan University of Science & Engineering, Zigong, Sichuan 644004, China;Key Laboratory of Higher Education of Sichuan Province for Enterprise Informationization and Internet of Things, Sichuan University of Science & Engineering, Zigong, Sichuan 644004, China)

机构地区：[1]四川轻化工大学自动化与信息工程学院,四川自贡644004 [2]四川轻化工大学计算机学院,四川自贡644004 [3]四川轻化工大学企业信息化与物联网测控技术四川省高校重点实验室,四川自贡644004

出　　处：《中国科技论文》2020年第3期300-307,共8页China Sciencepaper

基　　金：国家自然科学基金资助项目(61703409);四川省科技计划项目(2015JY0208);自贡市重点科技计划项目(2014DZ10,2019RKX03);企业信息化与物联网测控技术四川省高校重点实验室开放基金资助项目(2019WZY04);四川轻化工大学科研项目(2018RCL21)。

摘　　要：基于对工件螺孔形态的观察,分析了DXF格式的工件图纸文件,研究了文件管理控制芯片CH376的应用实例,提出了一种能有效纠正视觉误差、提高螺孔识别精度的方法。首先,将机器视觉技术与CAD技术相结合,在视觉识别的基础上,使用CH376枚举出用户U盘内的DXF文件,并根据DXF文件的构成规律查找出指定字段中的图元信息。然后,控制器利用图元依附关系、螺孔形状、螺钉国标规格等判定条件筛选出符合要求的坐标数据作为视觉误差纠正基准。最后,结合视觉与图纸数据,对实际工件中存在的平移偏差和旋转偏差进行纠正,演算出现场工件的正确螺孔坐标数据。实验结果表明,所提方法能够有效纠正视觉识别误差,且识别过程无需人工参与,工作速率和成功率均有明显提高。Based on the observation of the screw hole shape of the workpiece,the drawing interchange format(DXF)of the workpiece drawing file is analyzed.The application example of the file management control chip CH376 is studied.A method which could effectively correct the visual error and improve the screw hole recognition accuracy is proposed.First of all,combined machine vision technology with CAD technology,on the basis of visual recognition,CH376 was used to list the DXF files in user’s U disk and to find out the information of the specified fields according to the constitution of DXF files.Then,the controller selected the coordinate data that meet the requirements as the visual error correction benchmark by using the decision conditions such as entity attachment,screw hole shape,national screw standard specification and so on.Finally,combined with the vision and drawing data,the translation deviation and rotation deviation in the actual workpiece were corrected,and the correct screw hole coordinate data of the field workpiece were calculated.The experimental results show that the proposed method can correct the visual recognition error successfully.The recognition process does not need human participation,and the working rate and success rate are significantly improved.

关 键 词：文件管控芯片 图纸文件 纠正基准 平移偏差 旋转偏差 

分 类 号：TS736[轻工技术与工程—制浆造纸工程] TG659[金属学及工艺—金属切削加工及机床]