Edge Detection Based on a PCNN-Anisotropic Diffusion Synergetic Approach

A new synergetic model based on a pulse neural network and the Perona-Malik anisotropic diffusion algorithm is introduced. The proposed model was developed because of the failure of conventional edge detectors to perform the extraction of geometric structures from low contrast edges and noisy assumed uniform regions. The synergetic model is a variation of the Perona-Malik algorithm that incorporates a gradient update during its iteration process. The gradient update is related to enhance real edges and to reduce the noise content in uniform regions. The proposed approach is compared with the Canny and watershed methods as well as with a variation of the Perona-Malik algorithm. Results indicate that the synergetic approach yields a more complete geometric structure extraction compared with the others methods.