An adaptive multilevel thresholding method with chaotically-enhanced Rao algorithm

Area of Science:

• Computer Vision
• Image Processing
• Optimization Algorithms

Background:

• Multilevel image thresholding is crucial for image segmentation.
• Existing metaheuristic methods are complex and slow.
• Manual threshold determination limits practical application.

Purpose of the Study:

• To develop a simplified multilevel image thresholding approach using a novel optimization technique.
• To introduce Chaotic Enhanced Rao (CER) algorithms with automatic threshold number determination.
• To evaluate CER algorithm performance against established methods.

Main Methods:

• Development of Chaotic Enhanced Rao (CER) algorithms utilizing eight chaotic maps (Logistic, Sine, Sinusoidal, Gauss, Circle, Chebyshev, Singer, Tent).
• Automatic determination of the number of thresholds.
• Performance evaluation using statistical metrics (BDE, PRI, VOI, GCE, SSIM, FSIM, RMSE, PSNR, NK, AD, SC, MD, NAE) on the BSDS300 dataset.

Main Results:

• The proposed CER algorithm demonstrated superior performance in image segmentation based on PRI, SSIM, FSIM, PSNR, RMSE, AD, and NAE metrics.
• Experimental results confirmed the effectiveness of CER algorithms on the BSDS300 dataset.
• The method achieved better convergence speed and accuracy compared to existing approaches.

Conclusions:

• The developed CER algorithms offer an efficient and accurate solution for multilevel image thresholding.
• Automatic threshold number determination simplifies the segmentation process.
• The proposed method represents a significant advancement over traditional metaheuristic techniques.