Virus diffusion algorithm: A novel metaheuristic algorithm for segmenting the spun crack

Abstract Metaheuristic algorithms are extensively utilized in engineering due to their outstanding capacity for solving optimization problems with restricted computing resources or incomplete data. However, its extended use is constrained by the low optimization accuracy and premature convergence. The rapid spread and extensive reach of the COVID-19 virus have inspired the proposal of a new Virus Diffusion Algorithm (VDA) to overcome the limitations of the metaheuristic algorithm. This article utilizes the VDA algorithm to segment spun cracks, providing a method for intelligent detection of spinning process. The algorithm integrates global diffusion and local diffusion mechanisms to simulate both the random walk and local disturbance modes of virus diffusion, thereby enhancing its accuracy. Additionally, it introduces the competition mechanism and infection center rate to enhance the diversity of the population and expand the algorithm's search range. The effectiveness and robustness of the VDA algorithm is validated using the CEC'17 test benchmark function. Subsequently, the VDA algorithm is used to segment images with cracks in thin-walled spun parts. The experimentally obtained results illustrate that the VDA-based segmentation algorithm attains a PSNR of 23.6798 and an SSIM of 0.9864 for crack images, surpassing other segmentation algorithms in challenging conditions.