第30卷第1期2023年2月工程设计学报ChineseJournalofEngineeringDesignVol.30No.1Feb.2023基于表面评估的工件轮廓识别方法研究潘盛湖，谢林成，刘云强，徐尚飞（西南石油大学机电工程学院，四川成都610500）摘要：现有加工工件材料和加工方式的多样性使得工件的表面情况多样，导致视觉系统难以准确识别工件轮廓，因此提出了一种适用于不同工件表面的轮廓识别方法。根据纹理区域面积与凸包面积的比值对工件表面进行评估和分类。对于表面质量较好的工件，采用高通线性滤波器对图像进行滤波处理，通过工件表面信息与边缘信息的差异实现工件图像边缘提取；对于表面质量较差的工件，采用一种自适应轮廓提取方法来识别图像边缘。实验表明，与传统的Canny边缘检测算法相比，所提出的方法能够更好地去除噪声干扰，其识别轮廓的精度更高。所提出的轮廓识别方法对不同工件表面有较好的适应性，具有一定的应用价值。关键词：工件轮廓；表面评估；细化算法；边缘提取中图分类号：TP751文献标志码：A文章编号：1006-754X（2023）01-0013-07ResearchonworkpiececontourrecognitionmethodbasedonsurfaceevaluationPANSheng-hu,XIELin-cheng,LIUYun-qiang,XUShang-fei(SchoolofMechatronicsEngineering,SouthwestPetroleumUniversity,Chengdu610500,China)Abstract：Thediversityofexistingworkpiecematerialsandprocessingmethodsmakesthesurfaceoftheworkpiecediverse,whichmakesitdifficultforthevisionsystemtoaccuratelyrecognizetheworkpiececontour.Therefore,acontourrecognitionmethodsuitablefordifferentworkpiecesurfaceswasproposed.Theworkpiecesurfacewasevaluatedandclassifiedaccordingtotheratiooftextureareatoconvexhullarea.Fortheworkpiecewithgoodsurfacequality,thehigh-passlinearfilterwasusedtofiltertheimage,andtheedgeoftheworkpieceimagewasextractedbythedifferencebetweentheworkpiecesurfaceinformationandtheedgeinformation;fortheworkpiecewithpoorsurfacequality,anadaptivecontourextractionmethodwasadoptedtorecognizetheimageedge.ExperimentalresultsshowedthattheproposedmethodcouldbetterremovenoiseinterferencethanthetraditionalCannyedgedetectionalgorithm,anditscontourrecognitionaccuracywashigher.Theproposedcontourrecognitionmethodhasgoodadaptabilitytodifferentworkpiecesurfacesandhascertainapplica...
