HUSCAP logo Hokkaido Univ. logo

Hokkaido University Collection of Scholarly and Academic Papers >
Graduate School of Agriculture / Faculty of Agriculture >
Peer-reviewed Journal Articles, etc >

ハイパースペクトル画像解析によるカバークロップ圃場の構成草種および草量の空間分布推定

Files in This Item:
NK45-2_99-109.pdf1.24 MBPDFView/Open
Please use this identifier to cite or link to this item:http://hdl.handle.net/2115/50352

Title: ハイパースペクトル画像解析によるカバークロップ圃場の構成草種および草量の空間分布推定
Other Titles: Estimating Spatial Distribution of Herb Species and Herbage Mass in Cover Crop Field Using Hyperspectral Imaging
Authors: 鈴木, 由美子 Browse this author
岡本, 博史 Browse this author →KAKEN DB
平田, 聡之 Browse this author →KAKEN DB
片岡, 崇 Browse this author →KAKEN DB
柴田, 洋一 Browse this author
Keywords: remote sensing
image processing
spectral analysis
linear discriminant analysis
partial least squares regression analysis
Issue Date: 15-Jun-2010
Publisher: 日本農作業学会
Journal Title: 農作業研究
Volume: 45
Issue: 2
Start Page: 99
End Page: 109
Abstract: Cover crops have many benefits such as reducing chemical materials, improving soil condition, preventing soil erosion and conservation of soil water. The long-term objective of this study is to assess and estimate cover crop effectiveness such as green manure, weed depression and soil conservation. In this paper, spatial distribution of herb species and herbage mass was estimated. A ground based hyperspectral imaging, which is useful for acquiring field information, was employed to monitor the cover crop field (bristle oat and hairy vetch). In order to generate the maps of herbage mass, first, plant portions were extracted from hyperspectral images by NDVI (normalized difference vegetation index) threshold. Next, they were classified into plant species using linear discriminant models. Finally, the herbage mass of each plant species was estimated individually using partial least squares regression model, and mapped with gradient colors depending on the estimated value. The results show that the success rate of plant area extraction was 100% and the success rate of the plant species classification was 78.7%. With regards to the result of the herbage mass estimation, the model that used both plant pixel spectra and plant cover rate as explanatory variables had the highest and most stable accuracy than the other models. The maps of plant species and herbage mass reflected the actual spatial distribution on the field. It was demonstrated that the hyperspectral imaging system developed in this study is a useful technique for monitoring the cover crop.
Type: article
URI: http://hdl.handle.net/2115/50352
Appears in Collections:農学院・農学研究院 (Graduate School of Agriculture / Faculty of Agriculture) > 雑誌発表論文等 (Peer-reviewed Journal Articles, etc)

Submitter: 片岡 崇

Export metadata:

OAI-PMH ( junii2 , jpcoar )

MathJax is now OFF:


 

 - Hokkaido University