An improved edge direction histogram and edge orientation auto corrlogram for an efficient color image retrieval

This paper advances the edge histogram descriptor (EHD) of MPEG-7 standards and the edge orientation autocorrlogram (EOAC) by using the Full Range Autoregressive model (FRAR) with Bayesian approach. The model is applied in the HSV color space for extracting the edges. The proposed method preserves the chromatic information and the edges that are loosed by the spectral variations. The extracted edges are used to construct the EHD and EOAC. The proposed EHD and EOAC are extensively tested with the various image databases. Precision and recall method is used to measure the accuracy of the proposed method. The results evident that the proposed method outperforms the existing methods.