Multilevel Thresholding For Multispectral Images

Multilevel Thresholding For Multispectral Images

Normalized square difference based multilevel thresholding technique for multispectral images using leader slime mould algorithm

Authors: Manoj KumarNaik et al

Download link: https://doi.org/10.1016/j.jksuci.2020.10.030

Abstract: The existing methodologies used for multilevel thresholding are not efficient in terms of both accuracy and computation time. Two-dimensional histogram-based techniques are better in terms of accuracy while they are computation intensive. The slime mould algorithm used for optimization mainly depends on the best leader and two randomly pooled slime moulds from the population, which leads to poor exploitation with more iteration to converge. These problems are solved in this paper by introducing a novel normalized square difference (NSD) based multilevel thresholding technique using a leader slime mould algorithm (LSMA). The contributions are many fold – i) a NSD based multilevel thresholding method is proposed using the gray level and normalized square difference (GLNSD) 2-D histogram with reduced computation; ii) LSMA is suggested; iii) 23 classical and 6 modern composition test functions from the IEEE CEC 2014 test suite are considered for evaluation of LSMA; iv) experiments on multispectral images are presented. The benefits are – i) reduces computations, ii) improves accuracy. The qualitative metrics used for analysis include – search history, trajectory, and average fitness history. Scalability analysis and statistical analysis (using Friedman's mean rank test) are presented. The proposal is compared with state-of-the-art techniques and found better.

DOI: https://doi.org/10.1016/j.jksuci.2020.10.030

要查看或添加评论,请登录

Ajith Abraham的更多文章

社区洞察

其他会员也浏览了