Adaptive Noise Cancellation Using Recurrent Radial Basis Function Networks

摘要

The design of nonlinear Far Infrared (FIR) and Imaging Infrared (IIR) noisecancellation filters is discussed. Radial basis function neural network architectures are introduced. It is shown that by exploiting the duality with system identification that the nonlinear IIR filter can be configured as a recurrent radial basis function network. The network training, the inclusion of linear dynamic network links, and the metrics for performance monitoring are discussed. Examples to demonstrate the degree of noise suppression that can be achieved based on the new design are included.