Abstract:

In order to establish a method of subdividing young female' s torso shapes based on the morphology of longitudinal section curve，257 young college female students were selected. 3-D human body was acquired by scanner. Point cloud data of human body was simplified using reverse engineering software of Imageware. Point cloud data of longitudinal section profile curve was fitted by the cubic spline function, and subjected to wavelet denoising. Low frequency coefficient for for wavelet analysis was used to extract overall characteristic of signals. As for shape clustering the K-means cluster analysis was used, and the Davies-Bouldin was used to determine the optimal class number. Human body shapes could be divided into four types. The difference on all kinds of shapes on front/back center line, back, chest and hip were described. Finally, the prototype patterns of four types of body shape were built, and the relationships between patterns and body shapes were analysed, which provided the reference basis for the fitness design of young female' s patterns.

Key words: longitudinal section curve, wavelet denoising, wavelet decomposition, prototype pattern, young female body shape classification