一种利用高光谱像元分解技术提取水体边界的方法

针对水体边界混合像元导致的精度损失问题,提出了一种基于高光谱混合像元分解的水体边界提取方法。该方法结合高光谱影像水体边界混合像元特有的光谱特征,削弱诸多因素对水体边界像元识别的影响,获取水体边界混合像元,降低了混合像元分解的计算量。通过混合像元的高精度分解及水体边界像元分割,进一步逼近水体的真实边界,能显著提高水体边界界定的精度。试验结果表明:用该方法进行水体提取,精度明显优于水体指数法,略优于支持向量机法,总体精度为93.86%,Kappa系数为0.87。

A method of extracting water boundary using hyper spectral pixel decomposition technique

A water boundary extraction method based on hyper spectral mixed pixel decomposition is proposed in this paper to solve the problem of precision loss caused by water boundary mixed pixel.The method which is combined with the spectral characteristics of the hyper spectral images obtains the mixed pixels of the water boundary with weakening the impact of many factors on the pixel boundary recognition to reduce the computation amount of the mixed pixel decomposition.The precision of the boundary definition of water body can be significantly improved by approaching the true boundary of the water body with the high precision decomposition of mixed pixel and the segmentation of water boundary pixel.The experimental results show that the accuracy of the method is better than that of the water index method,which is slightly better than the support vector machine.The overall accuracy of this method is 93.86%,and the Kappa coefficient is 0.87.