Face Recognition: Pre-Processing Techniques for Linear Autoassociators

摘要

To improve the performance of a linear autoassociator, we explore the use of several pre-processing techniques: a Sobel operator, a Canny-Deriche operator, and a multiscale Canny-Deriche operator. The gist of our approach is to store, in addition to the original pattern, one or several pre-processed (i.e. filtered) versions of the patterns, here faces. We found that the multiscale Canny-Deriche operator gives the best performance of all models. In the framework of an automated face recognition system, we present also a module to be added prior to the neural network, based on Hough transform, for face location purposes.