Content Based Image Retrieval Using Improved Gabor Wavelet Transform and Linear Discriminant Analysis

Content based image retrieval (CBIR) system performs an important role in the retrieval of desired images from larger image database efficiently. It finds applications in all areas including government, hospitals, surveillance, architecture, journalism and many more. Improvement in the precision and recall of the CBIR systems is desired and challenging problem for all the researchers. CBIR system using collective color and texture feature extraction with linear discriminant analysis is proposed in this paper. Gabor filter and color histogram has proved to be very effective in describing visual content of the image. Images in database are described using concatenated feature vectors of texture and color. The computational complexity of obtaining the texture features through Gabor wavelet transform has been improved through down sampling of the query and database images. Linear discriminant analysis is a powerful and popular method to extract the exact information for retrieving the images from the database. It also assist in increasing recognition rate of the content based image retrieval even despite the fact that training set belongs to unsupervised nature. thus the combined features of texture and color obtained through Gabor wavelet transform and color histogram respectively with linear discriminant analysis assist in reducing retrieval time.