Publication Type:

Conference Paper

Source:

Computational Vision and Bio-Inspired Computing, Springer International Publishing, Cham (2020)

ISBN:

9783030372187

URL:

https://link.springer.com/chapter/10.1007%2F978-3-030-37218-7_38

Abstract:

Image segmentation is an activity of dividing an image into multiple segments. Thresholding is a typical step for analyzing image, recognizing the pattern, and computer vision. Threshold value can be calculated using histogram as well as using Gaussian mixture model. but those threshold values are not the exact solution to do the image segmentation. To overcome this problem and to find the exact threshold value, differential evolution algorithm is applied. Differential evolution is considered to be meta-heuristic search and useful in solving optimization problems. DE algorithms can be applied to process Image Segmentation by viewing it as an optimization problem. In this paper, Different Differential evolution (DE) algorithms are used to perform the image segmentation and their performance is compared in solving image segmentation. Both 2 class and 3-class segmentation is applied and the algorithm performance is analyzed. Experimental results shows that DE/best/1/bin algorithm out performs than the other variants of DE algorithms

Cite this Research Publication

V. SandhyaSree and Dr. Thangavelu S., “Performance Analysis of Differential Evolution Algorithm Variants in Solving Image Segmentation”, in Computational Vision and Bio-Inspired Computing, Cham, 2020.