决策树分类数码相机影像获取水稻覆盖度的研究

Study on Vegetation Coverage of Rice Used Photography Image Based on Decision Tree Classification Method

叶面积指数(LAI)、植被指数(VI)是利用高空间分辨率影像构建遥感长势监测、估产模型的重要参数,作物覆盖度可以间接表达叶面积指数和植被指数。高空间分辨率影像不仅能识别田块形状,提高农作物类型识别精度,还可以以田块为单元进行作物长势、单产监测。为此,快速、准确地获取地面作物覆盖度,建立覆盖度与叶面积指数、植被指数之间的定量关系,可为遥感监测提供一种重要参数。以2013年四川省德阳地区为研究区域,建立10个(1 km×1 km)样方。在水稻返青期(6月3日)、分蘖期(6月18日)、拔节期(7月3日)、抽穗期(7月25日)、灌浆期(8月12日)及成熟期(8月30日),以GPS定点,每样方采集5个地块,每地块3个重复,以数码相机垂直拍摄的方式,获取不同生育期的照片。利用ENVI 4.7图像处理软件,根据决策树分类方法,计算不同测点不同时期的覆盖度值。结果表明,此方法可准确、客观地提取水稻不同时期的覆盖度值;利用Photo Shop 6.0人工调节选择计算覆盖度的方法主观性较强,结果与决策树方法接近,无显著差异。

Vegetation index （Ⅵ） and leaf area index （LAI） are the parameters in the model of remote sensing monitoring of crop growth and crop yield estimation,which could be reflected by vegetation coverage （VC）.In terms of high spatial resolution image,the field shape can be sketched,and the precision of crop type identification improved in the field as a unit.To gain the vegetation coverage quickly and build the relationship between Ⅵ,LAI and VC is important.Ten quadrats （1 000 m × 1 000 m） in 2013,Deyang County,Sichuan Province were established.Five sample points were fixed with GPS per quadrat,and taken pictures three repeated each time every sample in the green returned stage （2013/6/3）,the tillering stage （2013/6/18）,the elongation stage （2013/7/3）,the heading stage （2013/7/25）,the grain filling stage （2013/8/12） and the harvest period （2013/8/30）.Based on ENVI 4.7 image processing software,the vegetation coverage in terms of decision tree method was calculated.It shown the result was reliable by the method.