A Scheme of Color Image Multithreshold Segmentation Based on Improved Moth-Flame Algorithm

IEEE Access

Jie Yu¹  Truong-Giang Ngo²  Trong-The Nguyen³  Hong-Jiang Wang³  Jeng-Shyang Pan³  Thi-Kien Dao³ 

[1] College of Mechanical and Automotive Engineering, Fujian University of Technology, Fuzhou, China;Faculty of Computer Science and Engineering, Thuyloi University, Hanoi, Vietnam;Fujian Provincial Key Laboratory of Big Data Mining and Applications, Fujian University of Technology, Fuzhou, China;

关键词: Moth-flame algorithm;    color image segmentation;    multi threshold segmentation;    minimum cross-entropy;   

DOI  :  10.1109/ACCESS.2020.3025833

来源: DOAJ

【 摘 要 】

A recently developed swarm intelligence algorithm by studying the natural moth's biological behavior is called Moth-Flame Optimization (MFO). The advantages of MFO conclude a simple structure and a robust selection capability. Still, it is easy to be trapped falling into optimal local, and slow search converges. This study suggests a new process improving MFO by hybridizing Lévy flight and logarithmic functions for its formula of flame updating to enhance the optimization performance of the algorithm. In the experimental section, a set of benchmark functions of CEC2013 and the multi threshold image segmentation are used to evaluate the proposed method performance. Compared results of the proposed methods with the different algorithms in the same condition scenarios show that the suggested approach provides better results than the various algorithms in the competitions.

【 授权许可】

Unknown