A Study on CNN Based Hole-Filler Template Design using Single-Term Haar Wavelet Series

The Cellular Neural Network (CNN) is an artificial neural network of the nearest neighbour interaction type. It has been used for image processing, pattern recognition, moving object detection, signal processing, augmented reality and etc. The cellular neural network CMOS array was implemented by Anguita et al [1 – 5] and Dalla Betta et al [6]. The design of a cellular neural network template is an important problem, and has received wide attention [7 – 9]. Based on the dynamic analysis of a cellular neural network, this paper presents, a design method for the template of the hole-filler used to improve the performance of the handwritten character recognition using single-term Haar wavelet series method (STHW). This is done by analyzing the features of the hole-filler template and the dynamic process of CNN and by using popular classical fourth order Runge-Kutta (RK) and STHW methods to obtain a set of inequalities satisfying its output characteristics as well as the parameter range of the hole-filler template. Some simulation results and comparisons are also presented.