Feature Selection Using Multi-Objective Modified Genetic Algorithm in Multimodal Biometric System

Today the multimodal biometric system has become a major area of study that is identified with applications of a large size in a recognition system. The feature selection is probably found to be the best factor to be optimized and is an on-going challenge in the midst of the optimization problems in the human recognition system. The feature selection aspires to bring down the number of the features, remove all types of redundant data and noise which result in a very high rate of recognition. The step further effects on the human recognition system and its performance. The work further presents a newer biometric system of verification that was multimodal and based on three different features which are the face, the hand vein, and the ear. This has today emerged as an extensively researched topic which spans various disciplines like signal processing, pattern recognition, and also computer vision. The features have been extracted by making use of the Incremental Principal Component Analysis (IPCA). Further, the work presented another novel algorithm of feature selection which was based on the Multi-Objective Modified Genetic Algorithm (MOM-GA). The Genetic Algorithm (GA) had been modified by means of introducing a levy search as opposed to a process of mutation. The algorithm has also proved to be an effective method of computation in which the search space is found to be highly dimensional. A classifier that makes use of the K-Nearest Neighbour (KNN) for classifying all accurate features is used. There were some investigations that were carried out and these results proved that this MOM-GA feature selection algorithm had been found as that which can generate certain excellent results using a minimal set of chosen features.