Modular Neural Networks for Human Recognition from Ear Images Compressed Using Wavelets

Abstract

This chapter is focused in the human recognition from ear images as biometric using modular neural networks with preprocessing ear images as network inputs [80]. We proposed modular neural network architecture composed of twelve modules, in order to simplify the problem making it smaller. Comparing with other biometrics, ear recognition has one of the best performances, even when it has not received much attention [9, 74]. To compare with other existing methods, we used the 2D Wavelet analysis with global thresholding method for compression, and Sugeno Measures and Winner-Takes-All as modular neural network integrator. Recognition results achieved was up to 97%.