As the most computationally intensive part of a stereo vision system, stereo matching has been the focus of intense research activities for the last four decades. We present the first attempts to compare a quantum stereo matching solution with state-of-the-art approaches on the Middlebury stereo datasets. We first looked at quantum annealing computation as a way to interact with the D-Wave quantum computer and then improved the quantum solution to the stereo matching problem found in the literature. Using a line-by-line approach, we traded accuracy off for the qubits availability in the QPU (Quantum Processor Unit). Our findings show that it is possible to obtain results from real-sized images despite the scarcity of physical qubits in the quantum hardware. While the current quantum solution solves a class P stereo matching problem, its real advantage over classical stereo matching algorithms will arise with NP-hard problems which will be the focus of our future research.
As the most computationally intensive part of a stereo vision system, stereo matching has been the focus of intense research activities for the last four decades. We present the first attempts to compare a quantum stereo matching solution with state-of-the-art approaches on the Middlebury stereo datasets. We first looked at quantum annealing computation as a way to interact with the D-Wave quantum computer and then improved the quantum solution to the stereo matching problem found in the literature. Using a line-by-line approach, we traded accuracy off for the qubits availability in the QPU (Quantum Processor Unit). Our findings show that it is possible to obtain results from real-sized images despite the scarcity of physical qubits in the quantum hardware. While the current quantum solution solves a class P stereo matching problem, its real advantage over classical stereo matching algorithms will arise with NP-hard problems which will be the focus of our future research.
Computer Vision (CV) labeling problems play a pivotal role in low-level vision. For decades, it has been known that these problems can be elegantly formulated as discrete energy-minimization problems derived from probabilistic graphical models such as Markov Random Fields (MRFs). Despite recent advances in MRF inference algorithms (such as graph-cut and message-passing methods), the resulting energy-minimization problems are generally viewed as intractable. The emergence of quantum computations, which offer the potential for faster solutions to certain problems than classical methods, has led to an increased interest in utilizing quantum properties to overcome intractable problems. Recently, there has also been a growing interest in Quantum Computer Vision (QCV), hoping to provide a credible alternative/assistant to deep learning solutions. This study investigates a new Quantum Annealing-based inference algorithm for CV discrete energy minimization problems. Our contribution is focused on Stereo Matching as a significant CV labeling problem. As a proof of concept, we also use a hybrid quantum–classical solver provided by D-Wave System to compare our results with the best classical inference algorithms in the literature. Our results show that Quantum Annealing can yield promising results for Stereo Matching problems, with improved accuracy on certain stereo images and competitive performance on others.
Computer Vision (CV) labelling algorithms play a pivotal role in the domain of low-level vision. For decades, it has been known that these problems can be elegantly formulated as discrete energy minimization problems derived from probabilistic graphical models (such as Markov Random Fields). Despite recent advances in inference algorithms (such as graph-cut and message-passing algorithms), the resulting energy minimization problems are generally viewed as intractable. The emergence of quantum computations, which offer the potential for faster solutions to certain problems than classical methods, has led to an increased interest in utilizing quantum properties to overcome intractable problems. Recently, there has also been a growing interest in Quantum Computer Vision (QCV), with the hope of providing a credible alternative or assistant to deep learning solutions in the field. This study investigates a new Quantum Annealing based inference algorithm for CV discrete energy minimization problems. Our contribution is focused on Stereo Matching as a significant CV labeling problem. As a proof of concept, we also use a hybrid quantum-classical solver provided by D-Wave System to compare our results with the best classical inference algorithms in the literature.
Quantum visual computing is advancing rapidly. This paper presents a new formulation for stereo matching with nonlinear regularizers and spatial pyramids on quantum annealers as a maximum a posteriori inference problem that minimizes the energy of a Markov Random Field. Our approach is hybrid (i.e., quantum-classical) and is compatible with modern D-Wave quantum annealers, i.e., it includes a quadratic unconstrained binary optimization (QUBO) objective. Previous quantum annealing techniques for stereo matching are limited to using linear regularizers, and thus, they do not exploit the fundamental advantages of the quantum computing paradigm in solving combinatorial optimization problems. In contrast, our method utilizes the full potential of quantum annealing for stereo matching, as nonlinear regularizers create optimization problems which are NP-hard. On the Middlebury benchmark, we achieve an improved root mean squared accuracy over the previous state of the art in quantum stereo matching of 2% and 22.5% when using different solvers.
In this paper, we propose a methodology to solve the stereo matching problem through quantum annealing optimization. Our proposal takes advantage of the existing Min-Cut/Max-Flow network formulation of computer vision problems. Based on this network formulation, we construct a quadratic pseudo-Boolean function and then optimize it through the use of the D-Wave quantum annealing technology. Experimental validation using two kinds of stereo pair of images, random dot stereograms and gray-scale, shows that our methodology is effective.
The quantum enhanced stereo vision technique is thus a groundbreaking effort toward autonomous robotics, where autonomous robots have high precision 3D mapping through the application of quantum computational principles to solve the problem of classical stereo vision systems. However, real-time robotic navigation and object recognition is not allowed by such traditional methods due to depth estimation errors, noise, and occlusion issues. Based on that, this research offers a Quantum Assisted Adaptive Stereo Vision System (QASVS) that makes use of quantum superposition to enhance the depth estimation parallelism, entanglement for noise reduction, and a quantum optimized disparity computation model to increase the mapping accuracy. The system makes significant improvements in depth perception and computing speed at a real-time adaptive level, which could not be accomplished by the above-mentioned methods. To test the proposed framework, it is tested in a dynamic environment and found to be performing better in low light and complex terrains. These systems can be most crucial in autonomous drones, robotic surgical systems, industrial automation, and space exploration because precise depth perception is necessary. Processing speed, image quality, and environmental adaptability are improved by comparing to traditional stereo vision techniques. The study foresees quantum computing as a revolution for 3D vision for robotics that in turn will lead to the quantum enhanced machine perception. The next step will be building quantum photonic processors, as well as applying AI-driven depth refinement for more optimization.
Quantum annealing is a form of quantum computing that exploits quantum effects to probabilistically solve a specific, NP-hard problem: finding the ground state of a classical, Ising-model Hamiltonian. Because physical quantum annealers are already available, there exists the pressing question of how to program such systems. That is, how can one map a computational problem into the coefficients of an Ising-model Hamiltonian for solution by quantum-annealing hardware? In this article, we address that question primarily from a practical standpoint. We survey extant software tools intended for programming D-wave annealing-based quantum processors and examine the programming model and solution technique promoted by each tool in an attempt to showcase the variety of contemporary approaches to solving computationally challenging problems on an existing annealing-based quantum computer.