Feature Extraction and Selection for Image Retrieval

In this paper feature extraction process is analyzed and a new set of edge features is proposed. A revised edge-based structural feature extraction approach is introduced. A principle feature selection algorithm is also proposed for new feature analysis and feature selection. The results of the PFA is tested and compared to the original feature set, random selections, as well as those from Principle Component Analysis and multivariate linear discriminant analysis. The experiments showed that the proposed features perform better than wavelet moment for image retrieval in a real world image database; and the feature selected by the proposed algorithm yields comparable results to original feature set, and better results than random sets.