Image Retrieval Based on Wavelet and Optimized Local Gaussian Difference Extrema Pattern

Abstract

An effective extraction of image features is very important in Content based image retrieval (CBIR). This paper presents a new CBIR scheme based on optimized combination of two texture features to increase retrieval performance. The proposed method uses wavelet and Local Gaussian Difference Extrema Pattern(LGDEP) as texture feature effectively. Performance of proposed system is formalized on different databases like: MIT-VisTex, Corel-lk. In addition, the experimentation carried for medical databases like: MESSIDOR (Retinal images), Magnetic Resonance Imaging (OASIS-MRI) database, VIA/I-ELCAP-CT lung database. The results shows substantial improvement in terms of average precision rate (APR) like: 92% to 99% for MIT-VisTex, 80% to 83% for MESSIDOR, etc.