Image Thresholding Segmentation Algorithm Based on Two-parameter Cumulative Residual Masi Entropy
Rong Lan, Lekang Zhang · 2022
In order to solve the problem that the thresholding algorithm based on Masi entropy is not effective in segmenting images with non-obvious grayscale changes, an image thresholding algorithm based on two-parameter cumulative residual Masi entropy is proposed in this paper. Firstly, the two-parameter Masi entropy formula is obtained by generalizing the classical Masi entropy with two adjustable parameters, and the flexibility of parameter selection improves the adaptability of the proposed algorithm to images; secondly, the concept of cumulative residual type entropy is used to propose a two-parameter cumulative residual Masi entropy formula to overcome the negative effect of non-obvious grayscale changes on image segmentation; finally, the two-parameter cumulative residual Masi entropy is used to construct the objective function, and the image segmentation is realized by maximizing the objective function. Experiments are conducted for nondestructive detection images and natural images, and the results show the effectiveness of the proposed algorithm.