中国农业科技导报 ›› 2019, Vol. 21 ›› Issue (2): 9-18.DOI: 10.13304/j.nykjdb.2018.0391

• 农业创新论坛 • 上一篇    下一篇

基于多视角图像的植物三维重建研究进展

周静静1,2,3,郭新宇2,3,吴升2,3,杜建军2,3,赵春江2,3*   

  1. 1.上海海洋大学信息学院, 上海 201306; 2.北京农业信息技术研究中心, 北京 100097; 3.北京市农林科学院数字植物北京市重点实验室, 北京 100097
  • 收稿日期:2018-07-02 出版日期:2019-01-15 发布日期:2018-09-05
  • 通讯作者: *通信作者:赵春江,中国工程院院士,博士生导师,主要从事农业信息化关键技术研究。E-mail: zhaocj@nercita.org.cn
  • 作者简介:周静静,硕士研究生,研究方向为图像处理、数字植物三维重建。E-mail: zjhnlywise@163.com。
  • 基金资助:
    国家自然科学基金项目(31671577);国家重点研发计划项目(2016YFD0300605-01);北京市农林科学院创新能力建设专项(KJCX20180423);北京市农林科学院数字植物科技创新团队项目(JNKYT201604)资助。

Research Progress on Plant Three-dimensional Reconstruction Based on Multi-view Images

ZHOU Jingjing1,2,3, GUO Xinyu2,3, WU Sheng2,3, DU Jianjun2,3,ZHAO Chunjiang2,3*   

  • Received:2018-07-02 Online:2019-01-15 Published:2018-09-05

摘要: 快速重建植株三维结构并以三维可视的方式分析研究农作物的形态结构和生长过程、进行表型测量是数字植物及作物育种研究的热点和难点。概述了植物三维重建的研究现状和主流植物三维重建技术,并对各种技术的方法原理和技术优势进行了分类和对比;重点介绍了基于多视角图像三维重建方法SFM和MVS的研究进展、技术原理以及相关软硬件系统平台;分析了近几年基于多视角图像三维重建方法在植物三维重建中的应用进展,基于多视角图像三维重建方法具有数据获取成本低、获取点云精度高、植物三维重建质量高等技术优势,同时指出了该技术方法的不足之处和未来技术发展趋势,以期为植物三维重建技术的发展、植物表型组学的研究以及低成本高通量设备的研制提供有益参考。

关键词: 三维重建, 植物表型, 运动恢复结构, 多视角立体视觉, 高通量表型

Abstract: Rapid reconstruction of plant three-dimensional structure and three-dimensional visual analysis of crop morphological structure, growth process and phenotype measurement are hot and difficult points in digital plant and crop breeding research. This paper summarized the research status and the mainstream technology of plant three-dimensional reconstruction, and classified and compared the methods, principles and technical advantages of various technologies. The research progress, technical principle and related hardware and software system platform of multi-view image three-dimensional reconstruction method SFM and MVS were emphatically introduced, and the application progress of multi-perspective image three-dimensional reconstruction in plant three-dimensional reconstruction in recent years were analyzed, the multi-perspective image three-dimensional reconstruction method had the advantages of low cost of data acquisition, high precision of point cloud acquisition and high quality of plant three-dimensional reconstruction. At the same time, the paper also pointed out the shortcomings of this technology and its future development trend, expected to provide useful reference for the development of plant three-dimensional reconstruction technology, the study of plant phenotypic omics and the development of low-cost and high-throughput equipment.

Key words: three-dimensional reconstruction, plant phenotype, structure from motion, multi-view stereo;high-throughput phenotype