Akademik Veri Yönetim Sistemi
11th International Conference on Advances in Statistics, Bologna, İtalya, 25 - 27 Nisan 2025, ss.29, (Özet Bildiri)
Image segmentation is the process of dividing an image into parts according to different features and making each part meaningful. Threshold-based segmentation is the most commonly used image segmentation method. These approaches are divided into two categories: bi-level and multi-level. The image is divided into two or more groups according to a threshold value by using threshold-based segmentation methods. In particular, they have been used extensively for grayscale images and are an advantage due to their low cost of computation. Multilevel thresholding, an extended version of bi-level thresholding, divides the image into many regions based on multiple thresholds. For this reason, multi-level thresholding is more appropriate for real-world problems. Otsu’s method, Kapur entropy, Masi entropy, Renyi entropy, and Tsallis entropy are the most common thresholding methods. However, when these methods are used, the segmentation performance decreases as the threshold level increases, and the computational cost rises exponentially. In order to overcome this problem, the use of metaheuristic algorithms has become very widespread. In this study, an opposition-based metaheuristic algorithm is proposed. The proposed meta-heuristic algorithm is developed using the objective functions of the Otsu and Kapur thresholding methods and applied to the medical images. It is compared with well-known metaheuristics to evaluate the performance of the proposed algorithm. Experimental results show that the proposed algorithm is superior to the other algorithms in terms of MSE, PSNR, SSIM, and FSIM as performance metrics.