北京航空航天大学学报 ›› 2020, Vol. 46 ›› Issue (2): 274-286.doi: 10.13700/j.bh.1001-5965.2019.0189

• 论文 • 上一篇    下一篇

基于三维点云模型的空间目标光学图像生成技术

陆婷婷1, 李潇1, 张尧1, 阎岩1, 杨卫东2   

  1. 1. 中国运载火箭技术研究院 研究发展部, 北京 100176;
    2. 河南工业大学 粮食信息处理与控制教育部重点实验室, 郑州 450001
  • 收稿日期:2019-04-28 发布日期:2020-03-11
  • 通讯作者: 陆婷婷 E-mail:tingtingspring@163.com
  • 作者简介:陆婷婷,女,博士,工程师。主要研究方向:视觉导航、图像处理、人工智能、指挥控制。

A technology for generation of space object optical image based on 3D point cloud model

LU Tingting1, LI Xiao1, ZHANG Yao1, YAN Yan1, YANG Weidong2   

  1. 1. Research and Development Department, China Academy of Launch Vehicle Technology, Beijing 100176, China;
    2. Key Laboratory of Grain Information Processing and Control, Ministry of Education, Henan University of Technology, Zhengzhou 450001, China
  • Received:2019-04-28 Published:2020-03-11

摘要: 空间探测任务中大量先验图像数据的缺乏,使得基于光学图像的态势感知和导航算法无法被有效定量测试和评估。针对此问题,提出了一种基于三维点云模型和射影变换基本理论的空间目标光学图像生成方法。在完成对空间目标三维点云模型和仿真摄像机模型构建基础之上,利用射影变换基本理论依次计算像平面所有像素点与空间目标三维点云模型空间点的对应关系,并基于Lambertian漫反射模型和相对应空间目标三维点云模型空间点的光照方向,得到所有像素点的灰度值,从而生成给定空间目标的光学图像。大量仿真实验表明:与传统的基于解析模型的仿真图像生成方法相比,所提的空间目标光学图像生成技术能够以更快的速度生成更加真实的仿真图像,且生成的仿真图像可以广泛应用于椭圆拟合、陨石坑检测、着陆器视觉导航、航天器交会对接、空间目标跟踪等典型空间应用算法的定性与定量评估。

关键词: 空间目标, 仿真图像, 点云模型, 射影变换, 人工智能

Abstract: The lack of the prior image data in the space exploration tasks makes it difficult to quantitatively test and evaluate the situation awareness and navigation algorithms based on the optical images. Accordingly, in this paper, we present an algorithm for generating the synthetic space object optical image based on the 3D point cloud model and the basic theory of the projective transformation. First, the 3D point cloud model of the space object and the optical camera model were constructed. Then, the corresponding pairs between all the pixels in the image plane and the space points of the 3D point cloud model were obtained via the basic theory of projective transformation, and subsequently the intensity of each pixel in the image plane was calculated by the lighting direction of its corresponding space point and the Lambertian reflection model, and finally the simulated image was generated. A great deal of simulation experiments demonstrate that the proposed algorithm can produce the more vivid simulated images rapidly than the traditional analytical image generation algorithm, and the generated images can be applied to testing and evaluating the typical space application algorithms qualitatively and quantitatively, such as ellipse fitting, crater detection, optical navigation landing on the planet, automated rendezvous and docking of spacecraft, 3D tracking of spacecraft, and so on.

Key words: space object, simulated image, point cloud model, projective transformation, artificial intelligence

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发