The infection algorithm: an artificial epidemic approach for dense stereo correspondence

Area of Science:

• Computer Vision
• Artificial Intelligence
• Bio-inspired Computing

Background:

• Stereo image matching is crucial for 3D scene reconstruction but presents significant computational challenges.
• Existing methods often struggle with complexity and efficiency in depth information extraction.

Purpose of the Study:

• To introduce a novel bio-inspired algorithm for efficient stereo image matching.
• To leverage an artificial epidemic process for enhanced 3D depth information extraction.
• To enable realistic view synthesis through optimized depth data.

Main Methods:

• Development of the 'infection algorithm,' a bio-inspired approach simulating epidemic processes.
• Application of distributed rules that propagate across images to identify matching features.
• Exploitation of image content to selectively compute necessary 3D depth information.

Main Results:

• The infection algorithm successfully performs stereo image matching on real image pairs.
• Efficient extraction of essential 3D depth information was achieved, reducing computational load.
• Generated realistic reprojected images, demonstrating the algorithm's effectiveness in view synthesis.

Conclusions:

• The proposed infection algorithm offers a computationally efficient and effective solution for stereo image matching.
• Bio-inspired approaches can significantly advance computer vision tasks like 3D reconstruction and view synthesis.
• This method provides a promising direction for future research in intelligent image processing.