Edge Detection Using Image Entropy and Pulse-Coupled Neural Networks

This paper presents a novel edge detection algorithm demonstrated in the following code implementation. To execute the demo, open the code in MATLAB and run the "demo" function to visualize edge detection results. The algorithm typically processes one image in approximately 3 seconds on standard desktop computers. The core innovation lies in integrating image entropy calculations with pulse-coupled neural networks (PCNN). The implementation workflow includes: 1) Calculating local image entropy to identify regions with significant information content 2) Applying PCNN synchronization dynamics to enhance edge responses 3) Combining these outputs through adaptive thresholding. Key MATLAB functions include entropyfilt for local entropy computation and custom PCNN iteration loops. Our technical paper "Combining Image Entropy with Pulse-Coupled Neural Networks for Edge Detection" presented at the International Symposium on Image Processing (September 2010) details the algorithm's mathematical foundation and experimental validation. This hybrid approach demonstrates superior edge detection accuracy compared to conventional methods, enabling more precise image analysis and information extraction through improved boundary identification.