Surface Quality And Efficient Detection Method Research Of Industrial Parts

With the expansion of the scale of industrial production,the demand for industrial components in all walks of life are also constantly expanding,and how to effectively detect industrial parts surface quality problem,will be our research topic.Through the understanding of the past parts detection methods,they have some problems of low detection efficiency and high cost.So,in order to solve the problem of industrial parts surface quality detection,this paper proposes a new method of description.Different from the previous methods of visual inspection of the surface quality of parts,this paper will use 3d point cloud data to complete the subject,the main contents of which are as follows:(1)This article proposes a new description method of 3d point cloud based on coefficient of spherical harmonic basis function description method of 3d point cloud,the method based on Legendre polynomial spread out the point cloud of generalized Fourier series,the resulting point cloud on the base of vector space coefficient of complete 3d point cloud of new descriptors,and then use this descriptor to represent point cloud model.(2)The 3d model of the industrial parts under test and the 3d model in the standard library are using this descriptor to represent,by comparing the calculation of Euclidean distance,and the similarity degree,the two models directly use the similar degree effectively,measurement under test and the difference of standard parts,and testing finished products.(3)In order to further effectively detect the defect position in industrial components,so this paper will be based on the coefficient of spherical harmonic basis function description method of 3d point cloud and three-dimensional point cloud segmentation method,combining with identified by defective parts 3d point cloud model,the method of using 3d point cloud segmentation to split it into each face,find the defect position compare similarity.Finally,this paper verifies the detection rate of this method by using the obtained 3d model data of industrial parts.For qualified parts,the detection rate reaches 100%,the detection rate of missing and deformed parts remains above 97%.