Region based Image Similarity using Fuzzy based SIFT Matching
Shashikanth CC, Parag A. Kulkarni · International Journal of Computer Applications · 2013
This paper proposes a region based approach using Scale Invariant Feature Transform (SIFT) and fuzzy logic for computing region based image similarity.When SIFT algorithm is used for matching images, a meaningful or semantic match according to human perception is not obtained.Hence to refine its output, a region based approach has been proposed.When a test image is compared with reference image, SIFT descriptors are computed and the images are segmented into regions and labeled.SIFT similarity measure along with the region information is given as input to fuzzy logic to determine region based similarity measure.Experiments are done using real world optical images, Caltech image datasets and a few satellite images.The proposed approach is found to be efficient for optical images, Caltech image datasets and have good performance for satellite images.