An efficient fuzzy based approach for color image retrieval with dual tree — Rotated complex wavelet and fuzzy histogram using selective relevance feedback approach

A rapid growth of digital information due to the vast development of internet and the availability of advanced digital devices has led to the huge development of digital information creating the challenges in various digital search applications in several areas such as medicine, commerce, education, and crime prevention. It is required to develop new content based search system low computational cost and enhanced retrieval accuracy. In this paper, we have presented an efficient and effective retrieval method based on Dual Tree-Rotated Complex Wavelet Filter and fuzzy histogram using fuzzy similarity approach. The retrieval is carried out by decomposing the images in the database using the wavelet and extracting texture features of the sub bands as feature vector components. The color features are obtained with fuzzy histogram using histogram linking method. The retrieval results are further improved by implementing fixed weight selective relevance feedback approach that selects and trains only the images with poor retrieval accuracy. The retrieval results obtained using selective relevance feedback approach is compared with the retrieval results of relevance feedback approach which trains all the images in database for computation time and complexity. The experimental results indicate that retrieval with selective relevance feedback provide an improvement in average retrieval accuracy of around 14% over retrieval without feedback with a reduction of approximately 70% less time as compared to retrieval results obtained with the relevance feedback approach.