A gradient-based method for multilevel thresholding

作者：

Highlights：

• A novel method locating multi-threshold points for image segmentation is proposed.

• The method is based on gradient decent and designed for discrete optimization.

• Performance of the method is verified by comparing with bionic algorithms.

• The method is more efficient than bionic algorithms.

摘要

•A novel method locating multi-threshold points for image segmentation is proposed.•The method is based on gradient decent and designed for discrete optimization.•Performance of the method is verified by comparing with bionic algorithms.•The method is more efficient than bionic algorithms.

论文关键词：Multilevel thresholding,Image segmentation,Gradient descent method,Kapur’s entropy,Otsu’s threshold

论文评审过程：Received 31 October 2019, Revised 21 January 2021, Accepted 2 March 2021, Available online 10 March 2021, Version of Record 23 March 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.114845