Study on extraction and curve fitting of arm-root contour in elderly women
Baixue Zhou, Jianmei Sun, Tong Yao, Jun Wang, Yan Liu, Li PanPurpose
Human body shape feature extraction and curve fitting are important research directions in the field of apparel fit and pattern generation, and this paper investigates the method of extracting arm-root contour points and curve fitting for elderly women. Aiming at the point cloud data of elderly women, the method of automatic extraction of feature points and segmentation of arm-root cross-section is investigated, and the arm-root curve model is established after further obtaining the arm-root contour point data.
Design/methodology/approach
Firstly, the point cloud data of 245 elderly women were obtained by 3D body scanning, and the arm-root feature points were automatically extracted by applying a-shape algorithm, slope rate of change, normal vector angle and other methods, and automatic segmentation of the human body point cloud was realized. Then, the fitting contour points were selected by centroid angle of polar coordinates, voxel grid downsampling and other methods. Finally, the arm-root curve model was obtained by curve fitting the contour points with a cubic B-spline curve.
Findings
The research results show that the method of extracting points according to point cloud geometric features can effectively realize the automatic extraction and segmentation of feature points of human point cloud data and obtain the arm root section. At the same time, the arm root curve model for extracting arm root contour points is good, and the fitting curve R2 test value is 0.99, the root mean square error is 0.29, and the maximum deviation is 0.64. The results of the study provide a basis for research on human body modeling and pattern generation for older women.
Originality/value
This method demonstrates excellent performance in extracting the arm root contour for elderly women, and the application of cubic B-spline curves achieves effective fitting of the arm root curve. Consequently, the proposed approach contributes novel insights to research on automatic feature point extraction in point cloud human models, automatic sectional contour extraction for occluded areas and curve fitting for other regions. It provides a data foundation for constructing comprehensive digital human models and generating garment patterns.