AI-assisted Single-Image Full-Frame Camera Calibration for Space-Constrained Stereoscopic Systems
Authors:
- Adithya SubramanyaASAdithya Subramanya
- Ram GopalRGRam Gopal
- Vandana SharmaVSVandana Sharma
Preprint
- 20 February 2024
- preprint
- Published by Optica Publishing Group in Optica Open
Abstract
Camera calibration plays a fundamental role in the wake of the newly emerging image data-driven technologies, where pinpoint accuracy in data is vital to the successful functioning of these systems. Conventional camera calibration algorithms require manual object placement, a process that can be exceptionally time-consuming and labor-intensive, particularly in scenarios where space constraints or delicate equipment are involved. We present an innovative calibration object and AI-aided pre-calibration routine with a specific emphasis on space-restricted environments. The proposed methodology obviates the need for manual multi-image acquisition. This is achieved by fabricating the novel calibration object, which contains 20 checkerboards in different positions and orientations. The precursor routine, assisted by an AI model, isolates and processes the individual checkerboards, which is then used as input for the camera calibration. We report an accuracy of 99.92% for ML-assisted checkerboard separation, with procedure time improved by nearly 64x and overall corrected reprojection error consistently below 0.5 pixels. Incorporating the proposed calibration routine into a 3D vascular imaging stereovision system, we demonstrate a depth resolution of 0.5mm.Keywords
This publication has 19 references indexed in Scilit:
- Robot–Camera Calibration in Tightly Constrained Environment Using Interactive PerceptionIEEE Transactions on Robotics, 2023
- Evaluation of a calibration rig for stereo laparoscopesMedical Physics, 2023
- Self‐calibration method of binocular vision based on conformal geometric algebraMathematical Methods in the Applied Sciences, 2022
- Reconstruction of 3D surfaces from incomplete digitisations using statistical shape models for manufacturing processesJournal of Intelligent Manufacturing, 2022
- Multi-view stereo in the Deep Learning Era: A comprehensive reviewDisplays, 2021
- Security and Privacy of Smart Home Systems Based on the Internet of Things and Stereo Matching AlgorithmsIEEE Internet of Things Journal, 2019
- Stereoscopic augmented reality in visual interface for flight controlAerospace Science and Technology, 2014
- Review of stereo vision algorithms and their suitability for resource-limited systemsJournal of Real-Time Image Processing, 2013
- A flexible new technique for camera calibrationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2000
- A versatile camera calibration technique for high-accuracy 3D machine vision metrology using off-the-shelf TV cameras and lensesIEEE Journal on Robotics and Automation, 1987