Efficient face recognition using morphological operations

There are many applications in which morphological operations play an important role. Morphology is a very useful mathematical tool or method to be used in image processing. It has many application like edge detection and extraction form images for specific purpose, feature extraction based on the boundary information etc. It has many advantages over other image processing tools like the complete outcome of the method lies in the selection of Structure Element(SE) and sequence of operations to be used. In face recognition also it has been used but only for binary images using the square shaped SE. Compensation for hair growth and cut have been assumed to be uniform which are not in real. There are also issues in image processing like corruption, occlusion, different facial expressions, poses which are tried to be solved by this method. A proposed methodology for a biometric system is described in this work. In this work the morphological operations are used for feature extraction in face biometrics while solving the above mentioned issues. In the preprocessing basic morphological operations are performed using circular SE for denoising and smoothing image. Feature extraction is done using the existing method of sparse representation based classification. Feature matching is done using eigenfaces and after getting initial results images are treated with more morphological transformations iteratively for getting better results. Here motto is to increase the efficiency of the biometric system for better results and less computation time using morphological operations. The morphological operations used here are based on the maximum and minimum intensity values so that they can be used for grayscale images. The system is tested on a customized database made from Indian face database (IIT K) and Essex University database.