Classification Of Young Female Breast Shape And Bra Size Recommendation

Breast shape is the basis of bra structure design,and the matching between bra and breast shape determines the fitness and comfort. The current national bra size standard only uses under-bust circumference and the difference between bust and under-bust circumference to divide sizes, and can’t reflect various breast shapes, which results in that consumers needs to try on many times to decide the bra size. Therefore, developing a bra size recommendation system based on breast shape classification can help customers to choose the appropriate bra size.This paper studied the breast cross section curve, sagittal section curve and bottom breast curve, classified breast shapes according to the average radius of curvature of the three curves, realized the quick breast shape recognition and automatic bra size recommendation. The paper provides a new method for the existing research of breast shape, and lays the foundation for bra personalized customization and virtual fitting.The following aspects of content are included in this paper:1. 266 young non-pregnant female students aged from 18 to 25 were selected as experimental objects, their point cloud data were obtained by 3D body scanning.2. Breast feature points and local breast coordinate system were defined by the Rapidform software. Breast cross section curve, sagittal section curve and bottom breast curve, which are in 3 dimensions were extracted according to the local coordinate system. The 9 average radius of curvature parameters of the 3 curves, which including longitudinal-upper, longitudinal-lower, longitudinal-global, transversal-inside, transversal-outside, transversal-global, bottom-inside, bottom-outside and bottom-global, were measured and regarded as the indexes describing breast shape.3. Simple commonly used body measurements which can be connected with bra pattern were selected and measured, the correlation between two kinds of parameters was analyzed, and regression models of average radius of curvature were established according to correlation analysis, the precision of the models were high.4. Average radius of curvature parameters were defined as classification index, a K-means clustering algorithm based on optimizing initial centers was used, pseudo F statistic was applied to determine the optimal number of clusters, breast shape was eventually divided into 7 clusters. A new bra size standard— "national bra size standard + breast shape cluster" was put forward, and the bra structure design for different breast shapes was proposed.5. Applied MATLAB software to build LVQ neural network breast shape recognition model, recognition precision reached 98% after training and testing the model.6. Visual Studio 2013 and NET Framework 4.0 were used to develop a young female bra size recommendation system, including 4 modules of recording users’ information and measurements, calculating the average radius of curvature, recommending bra size, showing breast shape curves. The evaluation results indicated that the system has convenient operation, thorough function, and can recommend bra size accurately.