Face Detection Using Boosting Algorithm And Ipsonet

Face Detection is a challenging research field in computer vision and is of great interest, as it is very first step in face processing system. In this paper, we introduced a popular learning method for solving classification problems i.e. AdaBoost, which combines number of weak classifiers into a strong classifier [4]. By taking AdaBoost classifiers in cascaded manner, a new boosting algorithm i.e. MLPBoost which we used as a strong classifier for face detection [15]. These Boosting algorithms are required to increase the speed and Accuracy of the detector, which is our prime goal. PCA i.e. Principal Component Analysis is a popular unsupervised statistical method to find useful image representation. This method finds set of basic images and represents faces as a linear combination of those images. . PCA is an efficient feature extraction method that we applied to improve an accuracy of detector. Another algorithm IPSONet i.e. Improved Particle Swarm Optimization [16] is used to reduce false positive images obtained after applying Boosting Algorithm. IPSONet is a training technique for neural networks like MLP i.e. multilayer perceptron that uses an improved PSO i.e. Particle Swarm Optimization to evolve simultaneously structure and weights of neural networks. Index Terms- AdaBoost, Face Detection, Particle Swarm Optimization, Principal Component Analysis.