CBIR using Multi-level DWT

In the Content Based Image Retrieval (CBIR) technique requires features such as color, texture, shape and hybrid features. In this work texture based feature is used to retrieve the images from the database. In the first instance, intensity component is separated from the color image. DWT is applied on intensity matrix. Maximum valued coefficients from LH, HL and HH are considered for feature vector. LL band is further decomposed to next three levels. Fourth level decomposed LL coefficients also be taken to construct the feature. Hence, feature vector contains all the resolutions of the image. Retrieval of the image is based on the distance between query and data base images. Experiments are performed on the Corel 500 Image Database. Finally, experimental results are evaluated by the recall, precision, and F-Score.