测绘通报 ›› 2019, Vol. 0 ›› Issue (8): 140-143.doi: 10.13474/j.cnki.11-2246.2019.0269

• 自然资源监测 • 上一篇    下一篇

结合光谱和纹理特征的林地变更检测

梅树红1, 范城城1, 廖永生2, 李雅然3, 施宇军1, 麦超1   

  1. 1. 广西壮族自治区遥感信息测绘院, 广西 南宁 530023;
    2. 广西壮族自治区地理国情监测院, 广西 南宁 530023;
    3. 南宁市国土测绘地理信息中心, 广西 南宁 530021
  • 收稿日期:2018-12-03 修回日期:2019-02-27 出版日期:2019-08-25 发布日期:2019-09-06
  • 通讯作者: 李雅然。E-mail:lyr19930813@163.com E-mail:lyr19930813@163.com
  • 作者简介:梅树红(1973-),女,硕士,高级工程师,主要研究方向为摄影测量与遥感。E-mail:meishuhong@163.com
  • 基金资助:
    广西创新驱动发展专项(桂科AA18118038);广西重点研发计划(2017AB54078)

Forestland change detection based on spectral and texture features

MEI Shuhong1, FAN Chengcheng1, LIAO Yongsheng2, LI Yaran3, SHI Yujun1, MAI Chao1   

  1. 1. Institute of the Guangxi Zhuang Autonomous Region Remote Sensing Information Surveying and Mapping, Nanning 530023, China;
    2. Institute of the Guangxi Zhuang Autonomous Region National Geographic Monitoring, Nanning 530023, China;
    3. Nanning Land Surveying, Mapping and Geoinformation Center, Nanning 530021, China
  • Received:2018-12-03 Revised:2019-02-27 Online:2019-08-25 Published:2019-09-06

摘要: 开展林地变更调查,能够为森林执法督察、林地"一张图"更新等提供精准的空间信息和属性信息,对于林业资源的监测管理具有重要意义。针对大范围多时相遥感影像人工勾绘变化图斑耗时费力的现状,提出一种结合光谱和纹理特征的林地变更检测方法,并以灵山县东北部为例,利用20171209和20180201两个时期的高分二号遥感影像进行试验。结果表明,该方法在减少人力投入、降低时间成本的基础上,不仅将遥感影像的变化检测效率提高了一半以上,同时能达到77%以上的检测准确率,在森林资源普查中具有一定的应用价值。

关键词: 主成分分析, 最大似然法, 归一化植被指数, 林地变更, 变化检测

Abstract: The investigation of forest land change can provide accurate spatial information and attribute information for forest law enforcement supervision and forest land "one map" renewal, which is of great significance for forest resources monitoring and management. In view of the time-consuming and laborious situation of large-scale multi-temporal remote sensing images, this paper presents a method of forest land change detection based on spectral and texture features. Taking the northeastern part of Lingshan County as an example, the GF-2 remote sensing images of 20171209 and 20180201 are used to carry out experiments. The results show that, on the basis of reducing manpower input and time cost, this method not only improves the detection efficiency of remote sensing image by more than half, but also achieves more than 77% detection accuracy. This method has certain application value in forest resources census.

Key words: principal component analysis, maximum likelihood method, normalized difference vegetation index, forestland change, change detection

中图分类号: