Research on Threshold Segmentation Method of Two-dimensional Otsu Image Based on Improved Sparrow Search Algorithm

Aiming at the issues of complex calculation and low accuracy two-dimensional (2D) Otsu segmentation images, an image threshold means 2D ground on a modified sparrow search algorithm is proposed. Firstly, in initialization stage, tent chaos mapping added to enhance multiformity population, population elite strategy introduced quality initial solution. Secondly, local reverse learning applied renewal location solve issue getting into optimality. Eventually, fused with segmented segmentation. Compared traditional algorithm, genetic (GA-Otsu), seagull optimization (SOA-Otsu), particle swarm (PSO-Otsu) (SSA-Otsu), mean square error (MSE) value reduced by 40.84%,2.68%, 1.57%,0.77% 1.04%, respectively, peak signal-to-noise ratio (PSNR) increased 24.48%, 1.24%, 0.83%, 0.40% 0.45%, respectively. Moreover, optimal proposed better than other five algorithms. It verified that this paper has faster convergence speed higher accuracy, effectively improves