Human Body Shape Clustering for Apparel Industry Using PCA-Based Probabilistic Neural Network

Aiming to cluster human body shapes much faster, accurately and intelligently for apparel industry, a new clustering approach was presented based on PCA_PNN model in this study, which referred to a kind of probabilistic neural network combining with principal components analysis. The specific implementation process could include the following steps. Firstly, the human body data were acquired by 3D anthropometric technique. Secondly, after being preprocessed, the human anthropometric data acquired were analyzed by PCA, in order to reduce the data volume. Sequentially, a PNN model for clustering human body shapes was established and evaluated, the input layer of which was composed of the factor scores resulting from PCA and the output layer of which was the body shape category. The experiment results showed that PCA_PNN model presented was a good nonlinear approximately network which was able to cluster male human body shape accurately and intelligently.