Asteroid shape reconstruction method based on convolutional neural network

Radar is the most powerful remote sensing tool for measuring the physical characteristics of near-Earth objects. In this paper, we propose an asteroid shape reconstruction method using convolutional neural networks, which can reconstruct the three-dimensional shape features of asteroids based on radar observation data. Compared to traditional methods, it has the advantages of automatic optimization and minimal manual intervention. Experimental results have been conducted to demonstrate the effectiveness of this method.