Multilevel thresholding of images with improved Otsu thresholding by black widow optimization algorithm

One of the most important methods of image processing is image thresholding, which is based on image histogram analysis. These methods analyze the image histogram diagram and try to present optimal values for the image thresholds so that the image regions can be distinguished by these thresholds. Thresholding is a popular method in image processing and is used in most research related to image segmentation due to its accuracy and efficiency. Multi-level thresholding, such as the Otsu method, is one of the most common methods of thresholding image processing. These methods have high computational complexity despite their accuracy and efficiency. When the number of thresholds used increases, these methods lose their efficiency due to increased complexity and execution time. One of the ways to find thresholds in the Otsu threshold method is to use metaheuristic algorithms such as the Black Widow Spider Optimization Algorithm. These algorithms can find the appropriate thresholds for the image at the logical time. In the proposed method, each threshold is a component or one dimension of a solution of the Black Widow Spider Optimization Algorithm, and an attempt is made to calculate the optimal threshold value without high complexity by this algorithm. Experiments on several standard images show that the proposed algorithm finds better thresholds than the particle swarm optimization algorithm, the firefly algorithm, the genetic algorithm, and the gray wolf optimization algorithm. The analysis shows that the proposed method in the PSNR index has a better value in 83.33% of the experiments than other algorithms and also in 80% of the experiments the proposed method has a better SSIM index than these methods. Analysis of the proposed algorithm on several pertussis images also shows that the proposed method has a good ability to threshold medical images such as brain tumors and optic disc detection in human retinal images.