Design for robustness counter detection CNN with applications

Abstract

This paper presents a theorem for designing the robustness template parameters of a cellular neural network (CNN) to detect contours in images. The theorem provides parameter inequalities for determining parameter intervals to implement corresponding tasks. As two first examples, the contour detection (CD) CNN has successfully detected contours in a grey pattern image and the Lena portrait. As the third example, the CD CNN has been used to analyze the characteristics of ultrasound B-scan images of six patients' livers. Using polynomials of order 10 approximates the data of the ultrasound B-scan images processed by the CD CNN. Primary analysis seems to display some relationships between the polynomial coefficients and the damage in the patients' livers