An Optimization Based Fast Multi-level Thresholding Technique

Otsu’s method is one of the best image thresholding techniques, but its complexity increases tremendously with the number of thresholds. In this work, Otsu’s criterion function is reduced into a suitable form for easy computation. An algorithm using Particle Swarm Optimization (PSO) is then proposed which works much faster than the original Otsu multi thresholding. This method achieves performance, at par with a similar approach using hybrid PSO, proposed recently, without going for the hybrid approach. It does so by suitable parameter selection for PSO and thereby avoiding some steps like sorting which are needed during the process of hybridization.