Content Based Image Retrieval Using Color Correlogram And Wavelet Transform

Abstract

Image processing is the course of exploit to distill the images by processing them. It covenants with image compression, noise removal, image identification, delineates gratitude, image retrieval and image variation. Image retrieval mea ns probing and retrieving the images from an image database, used in assorted meadows such as therapeutic, science, engineering etc. Customarily, there are two techniques, text-based image retrieval and content based image retrieval, used for getting the requisite images commencing the database. The text based image retrieval system, TBIR, assists to repossess the image from the database using annotations. The succession of multimedia technology and internet user cannot swear by means of the conventional techniques of information recuperation. CBIR extorts images from a hefty degree database using the visual contents of an input image also called low level features or features of an image. These visual features are extracted using feature extraction and then match with the input image to salvage the requisite image from the database. Histogram, color moment, color correlogram, gabor filter and wavelet transform are various CBIR techniques. These techniques can be used autonomously or pooled to acquire enhanced consequences. This work endows with a novel technique for retrieving the images from the image database using two low level features namely color and texture. The work introduces general color correlogram for color indexing and pyramid-structured wavelet transform for texture processing of the image. A formal discussion about the existing techniques is presented; furthermore the results of system s performance are illustrated. newline