Image retrieval with EMD for new perceptual color feature

This paper presents a novel scheme for perceptual Content-based Image Retrieval. Semantic information can be protected during the retrieval by means of new adaptive color quantization algorithm. The entire method is consisted of two steps: perceptual color feature development and comparing quantized color vectors. The experimental results show that the proposed scheme outperforms the traditional methods in terms of semantic retrieval and computing complexity. The groundtruth are made based on 60,000 images from Corel database.