For special highland lakes,the interference of snow and mountain shadows increase the difficulty of accurate extraction.In response to this deficiency,this paper takes Namco district as the study area,and selects TM images over Namco district in 2003,2009 and 2015.We propose a new method to extract plateau lake using the combination of the normalized water index and image segmentation technology.First of all,through the appropriate threshold selection,using NDWI to achieve the extraction of plateau lake information.Secondly,multi-scale segmentation and spectral difference segmentation are combined to ensure the integrity of the Namco Lake and the small objects are independent.Finally,by combining the result of water index method of the segmentation result,the complete Namco and surrounding lakes are extracted by voting principle.In order to quantitatively evaluate the method proposed in this paper,we use artificial visual interpretation method to sketch out Namco Lake area as a reference image.By using number of pixel errors criterion,compare the lakes extracted by NDWI-image segmentation method and lakes extracted by NDWI water index method only.The results show that the method of NDWI-image segmentation is a little more accurate.%对于特殊的高原湖泊来说,由于受到冰雪、山体阴影的干扰,增加了精确提取的难度.针对此不足,以西藏纳木错湖区为研究对象,分别选取2003年、2009年2期TM图像,提出了一种基于归一化水体指数和图像分割技术相结合的高原湖泊提取方法,进行纳木错湖区的提取.首先,通过合适的阈值选取,利用NDWI水体指数法实现高原湖泊信息的提取;其次,采用多尺度分割和光谱差异分割的图像分割方法相结合,保证了纳木错湖泊完整且细小地物独立;最后,将水体指数法的结果与分割结果有机结合,利用投票原理提取完整的纳木错及周围湖泊.为定量评价本文提出方法,以人工目视解译方法勾勒出的纳木错湖区作为参考图像,利用像元数量误差准则分别对NDWI-图像分割法提取的湖泊和仅利用NDWI水体指数法提取的湖泊作比较.结果显示,NDWI-图像分割的提取方法获得的精度稍高.
展开▼
