Robust content based image retrieval based on multi-resolution wavelet features and edge histogram

The paper presents an efficient and simple approach for content based image retrieval. It utilizes the strength of multi resolution wavelet transform and edge histogram to extract color and texture features. The wavelet based color feature extraction scheme performs better than existing Mpeg-7 color descriptors. Texture features are extracted from the edge histogram obtained from wavelet coefficients of the image at multiple resolution levels. The multi-resolution approach helps in collecting texture details from finer to coarser levels providing better performance than existing Mpeg-7 descriptors. The combination of the two features is trained and tested for Wang's image database. The results of retrieval are expressed in terms of precision and recall and compared with existing schemes. The results show the superiority of the scheme. The proposed scheme is fairly robust against several types of image alterations.