Hybrid Differential Evolution Using Low-Discrepancy Sequences for Image Segmentation

The image thresholding problem can be seen as a problem of optimization of an objective function. Many thresholding techniques have been proposed in the literature and the approximation of normalized histogram of an image by a mixture of Gaussian distributions is one of them. Typically, finding the parameters of Gaussian distributions leads to a nonlinear optimization problem, of which solution is computationally expensive and time-consuming. In this paper, an enhanced version of the classical differential evolution algorithm using low-discrepancy sequences and a local search, called LDE, is used to compute these parameters. Experimental results demonstrate the ability of the algorithm in finding optimal thresholds in case of multilevel thresholding.