Date of Award


Degree Type


Degree Name

Bachelor of Engineering Honours


Faculty of Communications, Health and Science

First Advisor

Dr Abdessalam Bouzerdoum


The study investigates the process of optimisation, implementation and comparison of a Shunting Inhibitory Cellular Neural Network (SICNN) for Edge Detection. Shunting inhibition is lateral inhibition where the inhibition function is nonlinear. Cellular Neural Networks are locally interconnected nonlinear, parallel networks which can exist as either discrete time or continuous networks. The name given to Cellular Neural Networks that use shunting inhibition as their nonlinear cell interactions are called Shunting Inhibitory Cellular Neural Networks. This project report examines some existing edge detectors and thresholding techniques. Then it describes the optimisation of the connection weight matrix for SICNN with Complementary Output Processing and SICNN with Division Output Processing. The parameter values of this optimisation as well as the thresholding methods studied are used in software implementation of the SICNN. This-two dimensional SICNN edge detector is then compared to some other common edge detectors, namely the Sobel and Canny detectors. It was found that the SICNN with complementary output processing performed as well or better than the two other detectors. The SICNN was also very flexible in being able to be easily modified to deal with different image conditions.