A differential evolutionary adaptive Harris hawks optimization for two dimensional practical Masi entropy-based multilevel image thresholding

Abstract

The prey gets completely exhausted in Harris hawks optimization (HHO), when the escape energy is zero. Its exploration ability is limited. The random operator selected in HHO is a waste of search agents (Harris hawks). To solve the problem, we propose a new differential evolutionary adaptive Harris Hawks optimization (DEAHHO). In this effort, the exploration phase is suitably modified to limit the escape energy within the range 2,0. Also, it is made adaptive to decide when the Harris hawk would do perch along with the other family members or move to a random tall tree. Further, the differential evolutionary concept is incorporated to enhance the exploration ability. Recently, 1-D histogram-based multilevel image thresholding methods, using Masi entropic function, are used for multilevel image thresholding. However, the contextual information is missing in 1-D formulation. To solve this problem, a new 2-D practical Masi entropy function is suggested in this work. Widely used 23 benchmark test functions are considered to validate our DEAHHO method. In this experiment, 500 images from the Berkeley Segmentation Dataset BSDS 500 are used. It is compared with the state-of-the-art methods and found better than other methods. The method could be used in the segmentation of biomedical images.

Keywords

• Machine intelligence
• Soft computing
• Optimal multilevel image thresholding