Threshold Segmentation of Magnetic Column Defect Image based on Artificial Fish Swarm Algorithm

Aiming at the low efficiency of magnetic column surface defect detection, vulnerability to human influence, and insufficient anti-noise performance existing 2D-OTSU threshold segmentation algorithm, an improved artificial fish swarm algorithm combined with was proposed improve accuracy real-time detection. Firstly, weight coefficient added on basis original distance function set optimize coefficient. The objective established by combining inter-class discrete matrix intra-class matrix, optimal obtained. Secondly, logistic model used perceptual range moving step size so as balance local global search ability convergence speed algorithm. Finally, is segment image, compared other algorithms four benchmark functions. Experimental results show that can effectively reduce time complexity At same time, for defects reaches 93%, which has good practicability.