A multi-level thresholding image segmentation algorithm based on equilibrium optimizer.

Area of Science:

• Computer Vision
• Image Processing
• Optimization Algorithms

Background:

• Multi-level thresholding is crucial for image segmentation but faces challenges with stability and accuracy for diverse image properties.
• Existing methods often struggle to find optimal thresholds for complex grayscale images.

Purpose of the Study:

• To propose a novel algorithm, the Advanced Equilibrium Optimizer (AEO), for determining optimal multi-level thresholds in grayscale image segmentation.
• To enhance the stability and performance of image segmentation using advanced optimization techniques.

Main Methods:

• The Advanced Equilibrium Optimizer (AEO) algorithm utilizes two sub-populations to balance exploration and exploitation.
• Two mutation schemes are incorporated to prevent local optima, and a repair function avoids duplicate thresholds.
• Performance is evaluated using benchmark images and metrics like cross-entropy, PSNR, SSIM, and FSIM.

Main Results:

• AEO demonstrated superior performance in multi-level threshold image segmentation compared to existing methods.
• The algorithm achieved high scores in cross-entropy, PSNR, SSIM, and FSIM, indicating effective segmentation.
• Experimental results confirmed the stability and accuracy of AEO in finding optimal thresholds.

Conclusions:

• The Advanced Equilibrium Optimizer (AEO) provides an effective solution for multi-level threshold image segmentation.
• AEO's unique approach enhances segmentation accuracy and stability across various image types.
• This method offers a significant advancement in image processing for achieving reliable segmentation results.