The robustness design of templates of CNN for detecting inner corners of objects in gray-scale images
Lei Ming, Lequan Min · 2004
The paper presents a theorem for designing the robustness template parameters of cellular neural/nonlinear network (CNN) for extracting inner corners of objects in gray-scale images. The theorem provides parameter inequalities for determining parameter intervals for implementing the corresponding tasks. The designed CNN has a linear A-template and a nonlinear B-template with two thresholds. A first numerical simulation example shows that the CNN designed via our method successfully detects the inner corners of objects in gray-scale images. A second one implies that the inner corner detection CNN may extract inner corners of objects in gray-scale images with Gaussian noise if suitable thresholds of the CNN are chosen.