Performance Study of the IceCube Upgrade Camera System

TL;DR

This paper evaluates the performance of the IceCube Upgrade Camera System, a novel calibration tool with nearly 2,000 cameras, using simulation-based image analysis methods including AI approaches to characterize Antarctic ice properties.

Contribution

It introduces a new calibration system with advanced image analysis techniques, including neural networks, for ice property measurement in the IceCube detector.

Findings

Preliminary results show effective image analysis performance.

AI-based methods outperform classical approaches in simulations.

The system can accurately determine optical module positions.

Abstract

The IceCube Upgrade Camera System is a novel calibration system designed to calibrate the IceCube detector by measuring the optical properties of the Antarctic ice. The system comprises nearly 2,000 cameras and illumination LEDs, which are present on every D-Egg and mDOM, the newly designed optical modules for the IceCube Upgrade. These units, deployed across the IceCube Upgrade volume, will capture transmission and reflection images that can be used to characterize the optical properties of both the refrozen ice within drill holes and the bulk ice between strings. Additionally, the images can aid in determining the positions of the optical modules the camera systems are mounted on. To maximize the systems performance, various image analysis methodologies have been explored, ranging from classical maximum likelihood estimation to AI-based approaches using neural networks. In this study, we present preliminary results on the performance of these methods based on images generated by a simulation tool developed specifically for this system.