地球信息科学学报 ›› 2015, Vol. 17 ›› Issue (4): 486-493.doi: 10.3724/SP.J.1047.2015.00486

• • 上一篇    下一篇

车载三维摄像仪数据的土地覆盖解译验证方法——以蒙古国东部为例

柏永青1,2(), 王卷乐1,3,*(), 陈义华4, 祝俊祥1,2   

  1. 1. 中国科学院地理科学与资源研究所 资源与环境信息系统国家重点实验室, 北京 100101
    2. 中国科学院大学,北京 100049
    3. 江苏省地理信息资源开发与利用协同创新中心, 南京 210023
    4. 合肥工业大学, 合肥 230009
  • 收稿日期:2014-08-28 修回日期:2014-11-30 出版日期:2015-04-10 发布日期:2015-04-10
  • 通讯作者: 王卷乐 E-mail:yongqingb@qq.com;wangjl@igsnrr.ac.cn
  • 作者简介:

    作者简介:柏永青(1992-),男,甘肃平凉人,硕士生,研究方向为农业信息化。E-mail:yongqingb@qq.com

  • 基金资助:
    中国科学院重点部署项目(KZZD-EW-08);中国科学院俄乌白人才专项项目;中国科学院信息化专项项目(XXH12504-1-01)

A Method of Land Cover Verification Based on 3D Camera: Taking Eastern Mongolia as an Example

BAI Yongqing1,2(), WANG Juanle1,3,*(), CHEN Yihua4, ZHU Junxiang1,2   

  1. 1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
    3. Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing 210023, China
    4. Hefei University of Technology, Hefei 230009, China
  • Received:2014-08-28 Revised:2014-11-30 Online:2015-04-10 Published:2015-04-10
  • Contact: WANG Juanle E-mail:yongqingb@qq.com;wangjl@igsnrr.ac.cn
  • About author:

    *The author: SHEN Jingwei, E-mail:jingweigis@163.com

摘要:

土地覆盖精度验证是定量刻画土地覆盖遥感解译数据质量的必要手段。传统的土地覆盖验证主要是以“GPS定点”+“野外景观照片”的平面验证方式,其获取的验证点数量受制于野外采样效率,且主观性强。本研究提出一种车载三维摄像仪的土地覆盖验证方法,解决了三维视频资料中GPS验证点信息的输出和提取、验证点分类识别与管理,打分的精度评价等技术问题。以蒙古国东部7省(市)采集的2013年野外视频为例,从264 GB的影像资料中,分别以1、5、10、60、300 s为采样间隔,分离和辨识出123 396、24 679、12 339、2056、411个验证数据,并以验证打分的方法获得本区域的验证精度。分析表明,车载三维摄像仪具有自动化程度较高、能够连续记录可观的三维实地信息的优点,适用于信息化和大数据时代的土地覆盖验证。实地分间隔抽样结果表明,60 s间隔的采样频率在验证效率上最优,变长间隔采样技术将成为未来需要深入研究的关键问题。

关键词: 土地覆盖, 遥感制图, 三维摄像仪, 精度验证, 蒙古国东部

Abstract:

Land cover mapping is an important aspect in remote sensing (RS) researches, and the precision validation is a necessary procedure to describe the quality of land cover data interpreted from RS images. GPS pointing combined with the field photos are mainly used in traditional validation methods, which is easily and subjectively affected by the field work efficiency and sampling place selection. A new land cover data precision validation method is proposed in this study, which is based on the vehicle-mounted three dimensional (3D) camera. A series of detailed technical issues are addressed, which include the outputs and extraction of the validated GPS points from 3D videos, the identification and management of these points, the precision evaluation technology based on the grading system, and so on. 264 GB videos were used, which were gathered in the field validation task from 7 provinces of eastern Mongolia in 2013. Taking 1, 5 and 10 seconds, and 1 and 5 minutes respectively as the sampling intervals, collecting at an average car speed of 50. 95 kilometers per hour, 123396, 24679, 12339, 2056 and 411 points were recognized accordingly. The precision validation results of the study area were obtained based on the scoring methods. The top score of overall accuracy is 71.01% with the 1 minute interval, the accuracies for meadow steppe, real steppe, desert steppe, built area, barren, cropland and water are 49.29%, 86.09%, 32.71%, 80.65%, 87.5%, 1 and 0, respectively. The analysis shows that the 3D videos are suitable for land cover validation due to its highly automatic and continuous working features in the information and Big Data era. The testing results of the field sampling with various temporal intervals indicate that 1 minute interval is optimal for the precision validation in eastern Mongolia. For improving the future studies, sampling with variable temporal interval is the key issue, and it should be advanced thoroughly for the video data automatic processing technology.

Key words: land cover, remote sensing mapping, 3D camera, precision validation, the eastern Mongolia