基于植被指数特征和空间分析的石漠化信息提取

Research on Extracting Rock-desertification Information Based on Vegetation Indexes and Spatial Analysis

通过对多光谱数据谱间特征与多种植被指数差异分析,采用改进的EVI指数提取石漠化信息,并使用ND-WI指数修正水体信息,结合空间坡度分析进行决策树分析,提高信息提取精度。以贵州省务川仡佬族苗族自治县CBERS-02B星CCD数据为例,研究最新的中巴资源一号02B星在采用植被指数特征和空间决策树分析对石漠化信息快速的准确性。研究表明这一方法可以克服只靠单一指数提取出现信息混淆问题,共抽查203个图斑,其中石漠化图斑179个,误判图斑19个,漏判的石漠化图斑5个,其总精度判率88.2%,是一种快速有效提取石漠化信息的方法。

In this paper, we adopt a new method which used improvement EVI index, NDWI index and Spatial-gradient analysis, to reduce the mixed information problems and ameliorate extracting technique, based on the analysis in Multi-spectral Remote Sensing Images and the varieties of main Vegetation Indexes. Using CBERS--02B CCD RS image data, it is researched on the principle and method of quickly extracting the Rock-desertification information in Wuchuan County Guizhou province. Experiments results show that our method drop off the problem of image information confusion. We spot-cheek 203 sampling vector polygons, including 179 rock-desertification vector polygans, 19 commission vector polygons, 5 omission vector polygons, and its overall accuracy reaches 88.2%.