Research And Design Of School Uniforms Batch Customization System Based On ISODATA Fuzzy Clustering

School uniform customization is a group oriented production for clothing with the development of mass production of MTM,its design and manufacture objects are students at school. School uniform customization is based on large scale, the cost is low and the quality is high while meeting specific needs. This paper on the basic of human body measurement and data analysis proposes a method using both of K-means and fuzzy ISODATA clustering, firstly it obtains an initial students body clustering using K-means, then obtains fuzzy clustering of all samples using ISODATA fuzzy clustering model, and thus designs the school uniform's size and layout of based on the results on body clustering of samples,.This paper's design method is to get a representative student's measurement data, by using principal component analysis to obtain main factor of student body, then it gets clustering results of samples after setting the initial clustering number with K-means method, finally, it uses the ISODATA fuzzy clustering model to obtain membership of clustering of every samples, which will further quantify the results of body clustering of samples and provides reliable quantity support for designing size and layout of school uniform.The specific realization of this paper is after collecting representative data from pupil samples, using principal component analysis to obtain main factor of body, then it integrates K-means and ISODATA fuzzy clustering model based on designed model of custom school uniform, by iterative running k-means and fuzzy ISODATA the two algorithms to obtain the clustering of samples' detailed quantitative results. The results of experiment find that under the premise of ensuring the collected data's accuracy, by setting the principal component analysis as the algorithm's input vector, then it uses the designed school uniform customization model to deal with the data samples, and obtains representative results of the clustering on samples, and the overall algorithm's result is also satisfied.