Unit panel nodes detection by CNN on FAST reflector

TL;DR

This paper presents a CNN-based method for detecting nodes on the FAST reflector in photogrammetry images, achieving high recognition accuracy and improving surface measurement precision.

Contribution

It introduces a CNN approach with candidate regions for node detection in reflector images, surpassing traditional edge detection methods.

Findings

Recognition rate of 91.5% with CNN-based detection

Significant improvement over traditional edge detection

Effective for photogrammetry of large radio telescope reflectors

Abstract

The 500-meter Aperture Spherical Radio Telescope(FAST) has an active reflector. During the observation, the reflector will be deformed into a paraboloid of 300-meters. To improve its surface accuracy, we propose a scheme for photogrammetry to measure the positions of 2226 nodes on the reflector. And the way to detect the nodes in the photos is the key problem in photogrammetry. This paper applies Convolutional Neural Network(CNN) with candidate regions to detect the nodes in the photos. The experiment results show a high recognition rate of 91.5%, which is much higher than the recognition rate of traditional edge detection.