植被重金属污染检测模型研究

Research on Detecting Model of Heavy Metal Pollution in Vegetation

利用高光谱遥感技术代替传统方法检测重金属污染,具有效率高、费用低、检测范围广等优点。但是高光谱影像的空间分辨率较低,为了提高精度需要提取影像的端元。鉴于纯净像元指数(Pixel Purity Index,PPI)法耗时长的缺点,提出一种基于高斯分布的波谱曲线概率法用于高光谱影像端元提取,并结合重金属胁迫下植被波谱响应变化建立了高光谱遥感影像的植被重金属污染检测模型。经过试验研究及分析,发现波谱曲线概率法端元提取的效果和精度与PPI相近,但是时间消耗明显减少。因此,建立的植被重金属污染检测模型可以用于高光谱遥感图像,具有一定的价值。

The use of hyperspectral remote sensing technology to replace traditional methods of detection of heavy metal pollution has advantages of high efficiency,low cost,wide detection range.But the spatial resolution of the hyperspectral images is lower,it＇ s needed to extract the endmember of the image to improve the accuracy.Given the pixel purity index （Pixel Purity Index,PPI） method is time-consuming,this paper presents a spectral curve probability method on hyperspectral image for endmember-extracting based on Gaussian distribution,and establishes a high vegetation remote sensing images of heavy metal contamination detection model combining with heavy metal stress changes in the spectral response of vegetation.Through experiments and analysis,it is found that the precision of spectral curve probability method is similar with PPI for endmember-extracting,but it＇ s time consuming significantly reduced and the heavy metal pollution detection model can be used to hyperspectral remote sensing image,it has a certain value.