Image Retrieval Based on Color and Shape Feature
YE Yin-lan · Journal of Taiyuan University of Technology · 2008
A new method of combining color,shape feature for efficient image retrieval is proposed.Color feature is extracted by the histogram of colors;shape analysis is done using wavelet moment.Then this paper implements similarity measurement of query image and image database by taking into consideration the similarity of both color and shape feature.In order to improve retrieval performance,a relevance feedback mechanism,based on support vector machines is proposed.Experiments show that the proposed system is not only efficient but also effective.