Cognition-based approach to parameter setting of directional difference filter using neural network based face recognition system

Directional difference filter (DDF) is a nonlinear filter for reducing impulse noise. It does not employ order statistics unlike median filter and can reduce impulse noise effectively. However, it has some parameters and we need to set them appropriately based on empirical control. So far, there are few studies to evaluate the appropriateness of the parameter setting of DDF. Parameter setting problem is generally solved by maximization or minimization of some objective evaluation functions such as correlation and statistical independence. However, it is not always easy to set such objective evaluation functions because we only have a single-channel noisy signal in DDF, and therefore it is difficult to evaluate whether the parameter is optimal or not. On the other hand, even when we cannot employ objective assumptions, we sometimes know that an image includes some subjective information such as face. Based on the above prospects, we directly use a neural network based facial recognition system as evaluation function for parameter setting of DDF. Experimental results show that cognition-based parameter setting has a potential to set adequate parameters of DDF.