Please wait a minute...
浙江大学学报（农业与生命科学版）
Journal of Zhejiang University (Agriculture and Life Sciences)  2018, Vol. 44 Issue (6): 765-774    DOI: 10.3785/j.issn.1008-9209.2017.11.101
    
Rice cropping information extraction mapping based on sample knowledge mining using high resolution remote sensing images
YIN Huafeng1, SU Cheng1, FENG Cunjun2, LI Yuqin1, HUANG Zhicai1, ZHANG Xiaocan1*
(1. Institute of Space Information and Technique, School of Earth Sciences, Zhejiang University, Hangzhou 310027, China; 2. Geomatics Center of Zhejiang, Hangzhou 310012, China)
Download: HTML ( )    PDF(11411KB)
Export: BibTeX | EndNote (RIS)      

Abstract  The rice fields in high resolution remote sensing images present mixed information constituted by distinct ground objects such as rice, soil, water, weed, duckweed and so on. Thus a novel approach for mapping of rice cropping areas based on sample knowledge mining was brought up according to spatial autocorrelation theory, which took advantage of the spectra combinational regularity. The accompanying mapping strategy was formulated based on this method. First, we segmented the high resolution remote sensing image into spectrally homogeneous base-units that represented distinct mixture information of several ground objects by grouping adjacent pixels with similar spectra. Second, we constructed a set of rice base-unit types through analysis of the base-unit types that contain rice field samples, and combine all the base-unit whose type belonged to this set to form initial rice cropping region. Finally, we vectorized the initial rice cropping region to initial rice cropping polygons, and then removed the polygons incompatible with spectra combinational regularity of rice fields through similarity analysis of combined feature of base-units between the rice cropping polygons and the rice field sample polygons. The overall accuracy of experimental rice cropping areas mapping results was over 96%. The successful application of this novel approach proves its efficiency and indicates its great potential for further utilization.

Key words： rice cropping information extraction      high resolution remote sensing images      sample knowledge mining      spatial autocorrelation
     
Published: 25 November 2018
CLC:  TP 79  
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Cite this article:

. Rice cropping information extraction mapping based on sample knowledge mining using high resolution remote sensing images. Journal of Zhejiang University (Agriculture and Life Sciences), 2018, 44(6): 765-774.

URL:

http://www.zjujournals.com/agr/10.3785/j.issn.1008-9209.2017.11.101     OR     http://www.zjujournals.com/agr/Y2018/V44/I6/765


基于样本知识挖掘的高分辨率遥感图像水稻种植信息提取方法

针对高分辨率遥感图像中水稻种植信息实际是水稻、泥土、水、杂草、浮萍等多种地物混合信息的情况，提出了一种基于样本知识挖掘的水稻种植信息提取方法。该方法以构成水稻种植信息的各种地物信息为分析的基本单元，依据空间自相关性理论，挖掘基于各种基本单元的水稻种植信息的组合特征，进而提出了一种水稻种植信息提取策略：首先，分割图像得到各类混合地物信息的基本单元；其次，通过分析水稻样本图斑所包含的基本单元种类确定构成水稻的基本单元类型，将相应类型的基本单元均归入初始水稻种植区；最后，通过分析矢量化的初始水稻种植区图斑内的基本单元的组合特征与水稻样本图斑内的基本单元的组合特征的相似性，剔除不符合水稻种植信息基本单元组合规律的初始水稻种植区；最后，通过分析矢量化的初始水稻种植区图斑内的基本单元的组合特征与水稻样本图斑内的基本单元的组合特征的相似性，剔除不符合水稻种植信息基本单元组合规律的初始水稻种植区图斑。实际的水稻种植信息提取结果表明，该方法实现了良好的提取效果，水稻提取总体精度可达96%。

关键词： 水稻种植信息提取,  高分辨率遥感图像,  样本知识挖掘,  空间自相关性 
[1] HU Gensheng, WU Wentian, LUO Juhua, HUANG Wenjiang, LIANG Dong, HUANG Linsheng.  Remote sensing monitoring of wheat aphids by combining HJ satellite images with least square twin support vector machine model.[J]. Journal of Zhejiang University (Agriculture and Life Sciences), 2017, 43(2): 211-219.