A QUBO Formulation of the Stereo Matching Problem for D-Wave Quantum Annealers

Area of Science:

• Computer Vision
• Quantum Computing
• Optimization

Background:

• Stereo matching is a fundamental problem in computer vision.
• Traditional methods often rely on complex optimization techniques.
• Quantum annealing offers a new paradigm for solving complex optimization problems.

Purpose of the Study:

• To propose a novel methodology for stereo matching using quantum annealing.
• To adapt existing computer vision formulations for quantum optimization.

Main Methods:

• Formulating stereo matching as a Min-Cut/Max-Flow problem.
• Constructing a quadratic pseudo-Boolean function from the network formulation.
• Optimizing the function using D-Wave quantum annealing technology.

Main Results:

• The proposed quantum annealing methodology effectively solves the stereo matching problem.
• Validation was performed on both random dot stereograms and gray-scale images.
• Experimental results demonstrate the efficacy of the approach.

Conclusions:

• Quantum annealing is a viable and effective approach for stereo matching.
• This work bridges computer vision and quantum computing for practical applications.
• The methodology shows promise for future advancements in 3D reconstruction.