Differential Evolution-Based Transient Search Optimizer for Image Multi-Thresholding Problem
Abstract A significant number of image processing and evaluation processes are presented and applied, due to their practical significance, in a variety of disciplines in the literature related to image processing and computer vision. One pre-processing approach that is frequently used to improve the...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | International Journal of Computational Intelligence Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44196-025-00821-8 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract A significant number of image processing and evaluation processes are presented and applied, due to their practical significance, in a variety of disciplines in the literature related to image processing and computer vision. One pre-processing approach that is frequently used to improve the information in a class of images is thresholding. By grouping correlated pixels according to the selected thresholds, the thresholding approach improves the image. For the benchmark image suite in this study, an entropy-based threshold is put into place. This study aims to investigate the thresholding performance of well-recognized fitness function called Kapur’s entropy, for a selected threshold. To facilitate the automatic identification of the optimum threshold (Th) on the benchmark images for a specified threshold value $$(Th=3, 5, 7)$$ ( T h = 3 , 5 , 7 ) , there is a need for an optimization algorithm that determines the optimum threshold values for the given image under study. This research proposed a self-adaptive hybrid differential evolution (DE) based transient search optimizer (TSO) called DETSO. DETSO uses the search equations of DE algorithm to improve the exploration capability of original TSO. Additionally, self-adaptivity has been included by modifying scaling factor and crossover rate using exponential decreasing and logarithmic decreasing mutation operator, respectively. The CEC 2019 numerical test problems have been used to confirm the working efficiency of DETSO. The experimental investigation confirms that utilizing a benchmark image test suite with varying dimensions, the DETSO helps to get superior results in achieving better threshold values in terms of performance metrics, such as PSNR, SSIM, FSIM, MSE, etc., compared to the other competitive heuristic algorithms. |
|---|---|
| ISSN: | 1875-6883 |