Abstract: In order to improve the efficiency and accuracy of intelligent automatic spraying of soles,the trajectory curve of automatic spraying of soles was planned and obtained based on point cloud processing technology. In the pretreatment of sole point clouds,straight-through filtering and Euclidean clustering were used to obtain sole body point clouds. An adaptive voxel down-sampling method for curvature feature division was proposed to analyze the curvature characteristics of shoe sole point clouds,introducing proportional error and Softsign function to adaptively obtain the voxel size,and remove redundant points while retaining features.In order to improve the accuracy of the glue spraying curve and solve the problem of sudden change in the contour curve of edge points obtained by scanning latitude and longitude lines,the contour of the highest point of the sole was reconstructed and optimized based on principal component analysis,Non-Uniform Rational B-Spline curve (NURBS) fitting was used to obtain the sole glue spraying trajectory curve. The test results show that the optimized and reconstructed sole gule spraying trajectory curve is smoother and more in line with the physical characteristics of the high edge of the sole,meeting the need for uniform glue spraying operation.

Key words: adaptive voxel downsampling, Principal Component Analysis (PCA), latitude and longitude scanning method, shoe sole glue spraying, point cloud