A new fusion of whale optimizer algorithm with Kapur's entropy for multi-threshold image segmentation: analysis and...

Area of Science:

• Computer Vision
• Artificial Intelligence
• Image Processing

Background:

• Image segmentation is crucial for object separation, but optimal thresholding remains challenging.
• Traditional thresholding methods are often slow, labor-intensive, and prone to inaccuracies.
• Multi-threshold segmentation of grayscale images requires efficient and reliable algorithms.

Purpose of the Study:

• To propose an Improved Whale Optimization Algorithm (IWOA) for accurate multi-threshold segmentation of grayscale images.
• To enhance the optimization process using linearly convergence increasing and local minima avoidance (LCMA) and a ranking-based updating method (RUM).
• To evaluate the performance of IWOA against seven other algorithms using standard metrics.

Main Methods:

• Developed an Improved Whale Optimization Algorithm (IWOA) incorporating Kapur's entropy for multi-threshold segmentation.
• Integrated LCMA to accelerate convergence and mitigate local minima issues during optimization.
• Implemented RUM to refine solutions and address limitations of the LCMA randomization process.
• Compared IWOA with seven existing algorithms on benchmark test images.

Main Results:

• IWOA demonstrated superior performance in multi-threshold image segmentation compared to other algorithms.
• The LCMA and RUM techniques effectively improved convergence speed and solution accuracy.
• Performance was validated using metrics including fitness values, PSNR, SSIM, Standard Deviation, and CPU time.

Conclusions:

• The proposed IWOA offers a computationally efficient and accurate solution for multi-threshold image segmentation.
• The integration of LCMA and RUM significantly enhances the optimization capabilities of the Whale Optimization Algorithm.
• IWOA presents a promising advancement for image segmentation tasks in computer vision and image processing.