An efficient ant-based edge detector

Abstract

An efficient ant-based edge detector is presented. It is based on the distribution of ants on an image, ants try to find possible edges by using a state transition function based on 5x5 edge structures. Visual comparisons show that the proposed method gives finer details and thinner edges at lesser computational times when compared to earlier ant-based approaches. When compared to standard edge detectors, it shows robustness to Gaussian and Salt & Pepper noise and provides finer details than others with same parameter set in both clear and noisy images. © 2010 Springer-Verlag Berlin Heidelberg.