A Crack Image Segmentation Algorithm Based on Adaptive T-distribution

Many researchers in the field of image segmentation have carried out a series of optimization on clustering algorithm, but in the case of complex crack segmentation, the clustering algorithm is particularly sensitive to noise and outliers, resulting in poor crack segmentation results. Therefore, a crack segmentation method based on adaptive T-distribution is used to improve the selection of clustering centers and make the cracks more accurate and complete segmentation. Firstly, the simulation verifies the improvement ability of introducing adaptive t-distribution mutation mechanism into Sparrow Search Algorithm (SSA) to avoid being caught in local optimization. Secondly, the optimized Sparrow Search Algorithm (TSSA) is used as initial clustering point of K-means algorithm. After verification, the proposed algorithm greatly improves the segmentation accuracy and fitness of complex crack images.