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

Abstract

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

Keywords

• Tent chaos mapping
• elite reverse learning strategy
• sparrow search algorithm (SSA)
• two-dimensional Otsu
• seagull optimization algorithm