A transit search algorithm with chaotic map and local escape operator for multi-level threshold segmentation of spleen CT images

Abstract

In response to the limitations of the Transit Search (TS) algorithm, such as slow convergence speed, susceptibility to local optima, and limited capability in handling complex optimization problems, this paper proposes an improved Transit Search algorithm based on chaotic map and a local escape operator (CLTS). The algorithm introduces chaos initialization in the initial stage to ensure more representative initial solutions. To enhance the efficiency of the transit phase, CLTS simplifies the process and improves the transit criteria with adaptive parameters. Additionally, CLTS introduces slime weight to improve convergence speed in the neighborhood phase. Finally, on the basis of the original exploitation stage, a local escape operator is introduced to effectively jump out of local optima and strike a balance between exploration and exploitation. Experimental results on both the CEC2017 and CEC2022 benchmark test sets demonstrate that the proposed CLTS algorithm achieves faster convergence speed and higher convergence accuracy compared to the TS algorithm and other advanced algorithms. Moreover, when combining CLTS with a denoising version of a multi-level threshold image segmentation model, it was applied for segmenting seven spleen CT images. The results indicate that CLTS is superior to the most advanced image segmentation algorithms in terms of convergence and segmentation effect.