There are many techniques for conducting image analysis and pattern recognition. This papers explores a way to optimize one of these techniques—image segmentation—with the help of a novel hybrid optimization algorithm. Image segmentation is mostly used for a semantic segmentation of images, and thresholding is one the most common techniques for performing this segmentation. Otsu’s and Kapur’s thresholding methods are two well-known approaches, both of which maximize the between-class variance and the entropy measure, respectively, in a gray image histogram. Both techniques were developed for bi-level thresholding. However, these techniques can be extended to multilevel image thresholding. For this to occur, a large number of iterations are required to account for exact threshold values. However, various optimization techniques have been used to overcome this drawback. In this study, a hybrid firefly and particle swarm optimization algorithm has been applied to yield optimum threshold values in multilevel image thresholding. The proposed method has been assessed by comparing it with four well-known optimization algorithms. The comprehensive experiments reveal that the proposed method achieves better results in term of fitness value, PSNR, SSIM, FSIM, and SD.

中文翻译：

---

一种应用于多级图像阈值的萤火虫和粒子群混合优化算法

有许多技术可用于进行图像分析和模式识别。本文探索了一种在新型混合优化算法的帮助下优化其中一种技术——图像分割的方法。图像分割主要用于图像的语义分割，阈值处理是执行这种分割的最常用技术之一。Otsu 和 Kapur 的阈值方法是两种众所周知的方法，它们分别在灰度图像直方图中最大化类间方差和熵度量。这两种技术都是为双层阈值开发的。然而，这些技术可以扩展到多级图像阈值。为此，需要大量的迭代来计算准确的阈值。然而，已经使用了各种优化技术来克服这个缺点。在这项研究中，混合萤火虫和粒子群优化算法已被应用于在多级图像阈值化中产生最佳阈值。通过将所提出的方法与四种众所周知的优化算法进行比较，对其进行了评估。综合实验表明，所提出的方法在适应值、PSNR、SSIM、FSIM和SD方面取得了更好的结果。