An optimal feature extraction technique for illuminant, rotation variant images

Extracting the features from images of various illuminations and rotations is a complex task. To overcome that, a novel image enhancement technique for extracting the optimal illuminant, rotation invariant features is proposed. Initially, preprocessing is performed by logarithmic transformation function which changes multiplicative illumination model in to additive one. Then NSCT based illuminant invariant feature extraction is applied. Inorder to reduce the size of the feature vector and to extract the useful information, a strong edge detector will be needed. Hence for feature selection, Ant colony Optimization algorithm is used. While applying this algorithm to the yaleB database, experimental results show that this algorithm yields the best subset of features. Also this integrated approach provides a better solution for complex illumination problems.