Battle royale optimizer for multilevel image thresholding

Area of Science:

• Computer Vision
• Image Processing
• Pattern Recognition

Background:

• Image segmentation partitions images into meaningful regions, vital for computer vision and medical imaging.
• Histogram-based thresholding, including Otsu's and Kapur's methods, is a common technique for image segmentation.
• Extending these methods to multilevel thresholding requires significant iterations, often necessitating optimization algorithms.

Purpose of the Study:

• To apply the Battle Royal Optimizer (BRO) for optimizing multilevel image thresholding.
• To evaluate BRO's effectiveness in image segmentation using the Berkeley segmentation dataset.
• To compare BRO's performance against other state-of-the-art optimization methods.

Main Methods:

• Multilevel image thresholding using the Battle Royal Optimizer (BRO).
• Segmentation of various images from the Berkeley segmentation dataset.
• Comparative analysis with existing optimization-based methods for image segmentation.

Main Results:

• BRO achieved optimal threshold values for multilevel image thresholding.
• The method demonstrated superior performance in terms of fitness value, Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Method (SSIM), Feature Similarity Index Method (FSIM), Color FSIM (FSIMc), and Standard Deviation.
• BRO outperformed other state-of-the-art optimization techniques.

Conclusions:

• The Battle Royal Optimizer (BRO) is a highly effective tool for multilevel image thresholding.
• BRO presents a promising and efficient solution for image segmentation tasks.
• This study highlights BRO's potential to advance image processing applications.