Image retrieval with visually prominent features using fuzzy set theoretic evaluation
Mayukh Banerjee, Malay K. Kundu, Prasanta Kr. Das · 2006
This paper proposes a new image retrieval scheme using visually significant features. Clusters of points around significant curvature regions (high, medium, weak type) are extracted to obtain a representative image. Illumination, viewpoint invariant color features are computed from those points for evaluating similarity between images. Relative importance of the features is evaluated using a fuzzy entropy based measure computed from relevant and irrelevant set of the retrieved images marked by the users. The performance of the system is tested using different set of examples from general purpose image database. Robustness of the system has also been shown when the images have undergone different transformations.