ABSTRACT
This paper proposes an algorithm named Histogram Mixture Model Genetic algorithm combining histograms and mixture models to segment images. The histogram of an image is approximated by several Gaussian functions with parameters. The parameters are obtained by the genetic algorithm with a fitness function defined by an error function. The algorithm improves the drawbacks of Otsu and non-parametric methods. In the experiments, the algorithm has better results than Matlab function graythresh for Otsu method as compared to the theoretical optimal value. The experimental results of moon images show that the hole in Marius Hills is segmented.
Disclosure statement
No potential conflict of interest was reported by the authors.