Multilevel thresholding using ant colony optimization

Thresholding is an important technique for image segmentation. The aim of an effective segmentation is to separate objects from the background and to differentiate pixels having nearby values for improving the contrast. The Otsu's method and Kittler's method are two of the most referred exhaustive thresholding methods. Our study proposes a hybrid optimization scheme based on an Ant Colony System algorithm with the Otsu and Kittler's methods respectively to render the optimal thresholding technique more applicable and effective. The ACS-Otsu and ACS-Kittler algorithms, two non-parametric and unsupervised methods, are the extension on the applications of the Ant Colony Optimization (ACO) for image segmentation. The experimental results show that ACS-Otsu algorithm outperforms ACS-Kittler algorithm in both CPU time and image quality in most level cases of test images.