Image Retrieval Based on Local Color and Edge Shape
Huilan Wang · 2009
The color moments of partitions are extracted from the L*a*b* color space to solve the problem that the global color moment excludes the color space distribution,the edge orientation autocorrelogram is used to represent the shape content of an image,and the feature distance is respectively computed.The Guassian model is used to normalize the feature.Based on the normalized distance,the global similarity measure is obtained,and the retrieval precision is improved by the weight value adjustment.Compared with the traditional local color mo-ment and the color moment retrieval,the method in the paper can obtain a better retrieval.