Shape Analysis Of Female Body Using 3D Body Scanning

The research of shape analysis of human body has significant value in the field of textile. Clothing can be designed better and more comfortable in the base of this research. Therefore, analysis and classify the morphological characteristics of main parts of human body can be a critical issue.In this paper, turning function and Ellipse Fourier Descriptors were used to analyze the shape variation of the 2D slices of female body. The 2D slice is the horizontal section of main parts of torso, including shoulders, chests, waist, abdomen, and hips. This thesis focuses on using the Fuzzy C-Mean Cluster(FCM) to classify the 2D slices, and using correlation analysis to validate the reliability of the results. The main contents of this paper lie in the following aspects:(1) Feature extraction. The feature points on 3D models should be defined at first, and then the feature planes can be obtained by segmenting with a horizontal plane. This thesis extracted the plane manually, and pretreated 3D data with Geomagic Studio. The segmentation is based on feature points, so the points should be identified clearly. The further research is based on the curves of cross-sections.(2) Feature slices fitting. The contours of cross-section are closed curves, and mostly similar to ellipse shape. Therefore, the Elliptical Fourier was processed to fit curves and calculated error, and identified the number of harmonic. The 2D slices pretreated, including noise reduction, center unity, normalization, and symmetrization, can be processed by fitting.(3) Cross-section shape analysis. Two methods, turning function and Elliptical Fourier, were used in this paper to analyze the shape. The contours of cross-sections were transformed into mathematical function image by the first method, and the variation of shapes can be reflected by the image. By the second method, the Elliptical Fourier Descriptors was calculated and the shape variation was analyzed using PCA. The results of the methods were compared.(4) Shape classification. This thesis classifies the cross-sections by cluster analysis, and the indicators were the main component scores. Verifying the effect of classification by correlation analysis, and the result can be reliable with high correlation coefficient.