Abstract: In this paper, Elliptic Fourier method was used to fit female’s bust, waist and hip slices, then principal component analysis was applied to the Elliptic Fourier descriptors. Multiple linear regression coefficients of principal components and slice’s width and depth were calculated to map principal components from slice’s width and depth. Slices were reconstructed from principal components and girths were derived. Fitting error of slice’s width, depth and girth of 459 samples were also analyzed. Bust slice has larger fitting error while waist slice has smaller. Finally, girth fitting errors caused by slice width and depth input error were shown in an regression form. The proposed method could be used in 2D photographic measuring system.

Key words: Elliptic Fourier, slice fitting, girth prediction, principal component analysis