90 Face Detection in Color Images Using Artificial Neural Networks

Image classification is a quantitative method that can be used to classify or identify objects or patterns on the basis of their multi-spectral values. Artificial Neural Networks (ANN) offer powerful solutions for pattern recognition. The Image classification can be achieved using neural networks because of its highly nonlinear properties. The objective of this paper is to create and train an artificial neural network, which will detect human faces in color images with neural networks using two methods. The networks discussed are based on: a) YIQ color space and b) XYZ color space. The two networks are trained using the same set of images by using the Levenberg-Marquardt Algorithm. The results obtained for the training set images and test images show that with a recognition rate of 97.57% and 97.23% for XYZ and YIQ approaches respectively, the face detection process using Artificial Neural Networks is successful.