A novel method of mapping semantic gap to classify natural images

Abstract

There exists an enormous gap between low-level visual feature and high-level semantic information, and the accuracy of content-based image classification and retrieval depends greatly on the description of low-level visual features. A novel method of mapping semantic gap is presented in this paper, which first extracts color and texture features from images after adaptive thresholding segmentation. Then, we use the BP neural network to map low-features to high-level semantic features. Experimental results show the efficacy of the proposed system.