Performance optimization of salp swarm algorithm for multi-threshold image segmentation: Comprehensive study of...

Area of Science:

• Computer Vision
• Artificial Intelligence
• Image Processing

Background:

• Multi-threshold image segmentation (MIS) is crucial for image analysis but often suffers from local optima with existing intelligent algorithms.
• The Salp Swarm Algorithm (SSA) is a metaheuristic optimization technique applied to various problems, including image segmentation.

Purpose of the Study:

• To propose an enhanced Salp Swarm Algorithm (EHSSA) to overcome the local optimum drawback in multi-threshold image segmentation.
• To improve the global search capability of the SSA for more effective image segmentation.

Main Methods:

• Developed a novel mechanism within the Salp Swarm Algorithm to enhance its global search efficiency.
• Applied the enhanced algorithm (EHSSA) to multi-threshold image segmentation tasks.
• Validated EHSSA performance through comparative experiments (IEEE CEC2014) and testing on the Berkeley Segmentation Data Set 500 (BSDS500).

Main Results:

• EHSSA demonstrated superior performance in multi-threshold image segmentation compared to existing methods.
• Experimental analysis confirmed the efficiency and improved global search capability of EHSSA.
• Successful application of EHSSA to microscopic breast cancer image segmentation, proving its effectiveness in medical image analysis.

Conclusions:

• The proposed EHSSA effectively addresses the local optimum problem in multi-threshold image segmentation.
• EHSSA offers a robust and efficient solution for both general and medical image segmentation tasks.
• The enhanced algorithm shows significant potential for improving diagnostic accuracy in medical image analysis.