Recognition of FAST reflector nodes based on Canny operator

Abstract

Abstract The reflector system of Five-hundred-meter Aperture Spherical radio Telescope (FAST) is designed as 4450 rigid panels supported by a flexible cable-net structure. We use 10 total stations to measure 2225 nodes of the cable-net and then control the shape of the reflectors. Every time, it takes at least 35 minutes to finish the calibration of the whole cable-net once. It is indeed far too inefficient. Thus, we developed a set of highly efficient instrument CRRS (CCD Rotation Ranging System). It is based on photogrammetry and can finish the measurement in 1 minute. However, the target we used in CRRS is active target, and it has serious electromagnetic interference problems to affect the use of FAST. Take the above reasons into consideration, we plan to identify the nodes by taking the gap between the reflector panels as the feature condition. The new method can take the place of active targets to finish the measuring task. The present work focuses on the following aspects. First, combined with the characteristics of FAST reflector images, the representative algorithms of image edge detection are discussed. Second, the process of node extraction is introduced in detail so that we know that it works. In addition, experimental results are given to draw a conclusion so that Canny algorithm was used for continuous research of reflector edge detection.