Image Retrieval Based on Color and Edge Distribution
Zhenxiang Shi · Computer Knowledge and Technology · 2010
A new retrieval method based on global color and shape (edge) features is proposed. This method modifies the gradient image generated by Sobel operator according the HSV Color Coherence Block, then describe shape feature using the Invariant Moments. And besides, Color information is described by HSV color quantification histogram in HSV color model. Results show that the average rates of precision of this method compared with RAHCB and common color quantification histogram is increased by 12.5% and 6.6% respectively.