A new algorithm for edge detection by hybrid differential evolution algorithm

In this paper, a new algorithm for edge detection was proposed. This method inspired by A. Bastürk's thoughts was formed, who proposed efficient edge detection using one neighbor CNN cloning template optimized by differential evolutionary algorithm. In order to consider interaction of more cells, and overcome solution's precocious phenomena, this paper extend one neighbor to two neighbors, and adopt hybrid differential evolutionary algorithm with a disturbance mutation operator optimizing two neighbors CNN cloning template. Through the general test images, simulation experiments indicate that the proposed method comparing with traditional edge detection methods has obvious advantage.