A Hierarchical Fuzzy Recognition Algorithm Based on feature Distribution Analysis
基于特征分布的图象信息抽取

In this paper,a new feature selection method is presented based on analysis of the feature distribution of the images using information theory.In order to extract more useful information of the feature distributions,we calculate the entropy of the feature intervals instead of the whole fearue space,then Special used Multi dimension Feature Space(SMFS) is constructed, which was comprise of effective feature intervals with small entropy, small inter pattern dispersion and large intra pattern dispersion.Different pattern has its distinctive SMFS.The experiment of the component recognition on the assembly lines using this method shows quite satisfing results.