Executive Summary : | Image segmentation is crucial in image processing for medical analysis, data mining, and pattern recognition. Traditional techniques face issues like local minima and premature convergence, and require considerable run time to find optimal pixels. To overcome these, a robust physics-based optimizer, Gravitational search Algorithm (GsA) and its variants will be employed for greyscale and color image segmentation. GsA is based on Newton's rule of global gravity and motion, and Kapur's entropy method is used to segment images based on pixel intensity values. The study will use benchmark images from the UsC-sIPI database, COVID-19 chest CT scan imaging datasets, X-ray images, and lupus nephritis images from the Kaggle database. Quantitative and qualitative metrics will be used to assess the quality, symmetry, and consistency of the segmented output. The signed Wilcoxon rank sum test will be applied to verify the simulation results. The main application of GsA is to provide faster and less computationally expensive segmentations, enabling early treatment and early diagnosis of COVID-19 patients. |