水库水体面积提取方法

Water area extraction method of reservoir

遥感影像的水库水体信息提取对水库面积变化监测有很大的帮助,因此,提出一种基于遗传算法和改进Otsu算法的水体提取方法。对处理后的遥感影像使用NDWI(normalized difference water index)水体指数法进行初始的水体提取,由于传统的Otsu算法对直方图呈现双峰分布的图像提取效果不佳,利用遗传算法对最大类间方差公式进行双阈值计算,引入滑动窗口对图像进行阈值判断;使用自适应阈值算法进行局部阈值分割。通过对石梁河水库和小塔山水库的实验,表明该方法能够准确提取出水库的水体信息,误提取和漏提取现象得到了很大的改善。

The extraction of water information from remote sensing image is of great help to the monitoring of reservoir area change.Therefore,a water extraction method based on genetic algorithm and improved Otsu algorithm was proposed.NDWI(normalized difference water index)water index method was used for initial water extraction of the processed remote sensing image.Since the traditional Otsu algorithm has poor extraction effects on the image with bimodal distribution of the histogram,the genetic algorithm was used for double threshold calculation of the maximum inter-category variance formula,and the sliding window was introduced to judge the threshold value of the image.The adaptive threshold algorithm was used for local threshold segmentation.Results of experiment of Shiliang River reservoir and Xiaotashan reservoir show that this method can accurately extract the water information of the reservoir,and the phenomenon of false extraction and missing extraction is greatly improved.