Integrating agent-based models and clustering methods for improving image segmentation.

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Clustering-based image segmentation is common but struggles with noise and variations in real-world images.
• Pixel feature homogeneity is often compromised, leading to inaccurate classifications.
• Existing methods lack robustness for complex image datasets.

Purpose of the Study:

• To develop a novel hybrid image segmentation method.
• To enhance segmentation accuracy and robustness using a combined approach.
• To overcome limitations of traditional clustering techniques in noisy environments.

Main Methods:

• A hybrid approach combining an agent-based model with Firefly metaheuristic clustering.
• Agent-based model preprocesses images by homogenizing pixel intensities via neighborhood consensus.
• Firefly clustering algorithm segments the preprocessed image into distinct regions.

Main Results:

• The hybrid method demonstrated superior performance on various test images.
• Enhanced image quality and robustness were observed compared to other methods.
• Key quality indices confirmed the effectiveness of the proposed approach.

Conclusions:

• The hybrid agent-based and Firefly clustering method offers improved image segmentation.
• This approach is robust and accurate, particularly for complex and noisy images.
• The technique advances the field of computer vision and image analysis.