A Novel Differential Exchange Parallel Great Wall Construction Algorithm for Efficient Multilevel Image Segmentation

The Great Wall Construction Algorithm (GWCA) is an innovative meta-heuristic algorithm for solving engineering challenges. Parallel strategies are an effective approach to enhancing the performance of algorithms. In this paper, parallel technology is introduced into GWCA by proposing a new communication strategy named Differential Exchange, which is combined with GWCA to design the Parallel Great Wall Construction Algorithm (PGWCA). The goal is to avoid local optima and improve the algorithm’s performance. The performance of the proposed PGWCA has been critically compared with seven other meta-heuristic algorithms on 10, 30, and 50 dimensions, respectively, using the CEC 2017 benchmark function suite. In addition, the PGWCA algorithm has been used to solve the multilevel image threshold segmentation problem by segmenting eight images and evaluating the segmentation results using three metrics: PSNR, SSIM, and FSIM. The results show that out of 29 test functions, PGWCA achieved the best (minimum) value 17, 20, and 16 times, respectively. In the image threshold segmentation experiments, the best performance was achieved 60 times out of all 72 evaluated comparisons. The experimental results show that PGWCA achieves better performance in identifying the optimal solution and improving the convergence performance, and has good performance in image segmentation. The code is available for download at the following link: https://drive.google.com/drive/folders/1I-5k-IzTP7cUvZqdo7YV-jOz1Zfka42j?usp=sharing