Face recognition system of infra-red images using ensemble back-propagation neural networks

Various kinds of Unmanned Aerial Vehicles (UAV) have been developed for many purposes, and numerous studies concerning with the development of components and its applications arises considerably. This paper addresses the development of a face recognition system to be incorporated into a UAV system for security application. We present an approach by using ensemble neural networks with negative correlation learning based on quadratic error function for recognizing unlearned face taken from a near infrared camera. Various experimental settings for studying the performances of the developed algorithms were conducted in three different scenarios; a conventional single component neural network with back-propagation learning algorithm, an ensemble neural network with two different learning methods, and an ensemble neural network trained by various percentage of learning dataset for both learning methods. Experiment results show that the recognition rate of the developed system could achieved as high as 99.9% for a database that consists of Indonesian persons.