On the use of a consumer-grade 360-degree camera as a radiometer for scientific applications

TL;DR

This paper demonstrates how to calibrate a consumer-grade 360-degree camera for scientific radiometry, validating its accuracy and potential for accessible, low-cost light measurement applications.

Contribution

The study develops calibration methodologies for consumer 360-degree cameras to enable their use as radiometers in scientific research.

Findings

Calibration accuracy with less than 21.1% bias compared to standard systems

Limit of detection of at least 4.6 x 10^-7 W·sr^-1·m^-2·nm^-1 in spectral channels

Validation of calibration through real-world sky radiance measurements

Abstract

Improved miniaturization capabilities for complex fisheye camera systems have recently led to the introduction of many compact 360-degree cameras on the consumer technology market. Designed primarily for recreational photography, several manufacturers have decided to allow users access to raw imagery for further editing flexibility, thereby offering data at sensor level that can be directly exploited for absolute-light quantification. In this study, we demonstrate methodologies to carefully calibrate a consumer-grade 360-degree camera for radiometry use. The methods include linearity analysis, geometric calibration, assessment of the illumination fall-off across the image plane, spectral-response determination, absolute spectral-radiance calibration, immersion factor determination and dark-frame analysis. Accuracy of the calibration was validated by a real-world experiment comparing sky radiance measurements with a co-localized Compact Optical Profiling System (C-OPS, Biospherical Instruments Inc.), which gave mean unbiased percentage differences of less than 21.1 %. Using the photon-transfer technique, we calculated that this camera consisting of two fisheyes with a 182$^\circ$ field of view in air (152$^\circ$ in water) has a limit of detection of at least $4.6 \times 10^{-7}$ $\mathrm{W \cdot sr^{-1} \cdot m^{-2} \cdot nm^{-1}}$ in its three spectral channels. This technology, with properly stored calibration data, may benefit researchers from multiple scientific areas interested in radiometric geometric light-field study. While some of these radiometric calibration methods are complex or costly, this work opens up possibilities for easy-to-use, inexpensive, and accessible radiance cameras.