Three-dimensional asteroid reconstruction algorithm based on Multilayer Perceptron
Ziyang Pan, Zhenming Tian, Chen Yan, Zegang Ding, Zehua Dong · 2024
The three-dimensional reconstruction of asteroids is one of the most popular fields within asteroid research. In this paper, a simple Multilayer Perceptron neural network is proposed, which relies on ISAR radar to reconstruct the three-dimensional point cloud model of the asteroid. Compared with the traditional algorithm, it does not need to manually adjust a lot of parameters to complete the reconstruction of different asteroids, and the network structure is simple and lightweight, requiring very little training data. Simulation results demonstrate the effectiveness of this algorithm.