基于混合像元分解的水体面积提取算法

为减小水体边界混合像元对水体面积提取精度的影响，提出基于混合像元分解的水体面积提取改进算法。以ALOS数据为例，利用混合像元线性分解模型获得各像元水体丰度，结合阈值法和数学形态学算法分别提取纯净水体像元和水体边界像元；通过分析水体边界像元的水体丰度构成，提出水体边界像元面积修正算法，实现水体面积修正。研究结果表明，与混合像元分解直接提取的水体面积相比，修正算法提取的水体面积精度最大提高6.7%，有效减小滩地水分对水体边界像元水体丰度计算的影响。

Water body area extraction algorithm based on pixel unmixing

An improved algorithm for water body area extraction based on pixel unmixing is proposed in order to reduce the impact of water body boundary pixels on water body area extraction. A fraction image of water abundance was obtained with the linear pixel unmixing model using ALOS data. The threshold method and the mathematical morphology method were used to extract pure water and water body boundary pixels. Through analysis of the composition of water body boundary pixels, an area correction algorithm for water body boundary pixels is proposed to correct the water body area. The research results show that, compared with the linear pixel unmixing model, the proposed method can improve the accuracy of water body area extraction by 6. 7% at most, and can reduce the impact of the water content of washlands on the calculation of water abundance of water body boundary pixels.