SPARSE REPRESENTATION AND COMPRESSION DISTANCE FOR FINDING IMAGE SIMILARITY

For the image similarity sparse representation is widely used because of it’s simplicity and easiness. Sparse representation and compression distance plays an important role in finding the similarity between the two images. For image similarity we create an overcomplete dictionary. Dictionary may be complete or over complete depending upon the elements contain in it. Overcomplete means the basic element or at oms in dictionary is greater than the vector space of that element. For the feature extraction we classify all the images into different classes and perform clustering on that so that it is easy to match the different images with the original one. Sparse R epresentation simply create a dictionary of the respective images and extract feature from it and perform matching on it.