改进RFLICM模型的青海湖水体边界提取  

Water Boundary Extraction of Qinghai Lake Based on Improved RFLICM Model

作　　者：张得琳 沈吉宝 ZHANG Delin;SHEN Jibao(Xining Land Survey and Planning Research Institute Co.,Ltd.,Xining 810000,China;Gansu Institute of Surveying and Mapping Engineering,Lanzhou 730000,China)

机构地区：[1]西宁市国土勘测规划研究院有限公司,青海西宁810000 [2]甘肃省测绘工程院,甘肃兰州730000

出　　处：《地理空间信息》2023年第10期54-57,共4页Geospatial Information

基　　金：省级科研立项课题(GSBSM-2020-01-02/4)。

摘　　要：针对传统FCM聚类方法提取青海湖水体边界的不确定性,提出改进模糊局部信息聚类方法(RFLICM)的青海湖水体边界提取方法。该方法利用条件概率密度函数的负对数描述像素与聚类间相似性,并结合空间位置定义模糊局部变量以控制算法分割尺度,结合邻域像素标号定义先验概率,进而将影像分割成湖水和陆地两部分,通过提取影像分割结果轮廓线获得湖水边界线。实验结果表明,该方法可以有效地提取青海湖边界信息,提取结果优于RFLICM与隐Markov随机场模型的增强型空间约束FCM影像聚类(HMRF-FCM)方法。Aiming at the uncertainty of Qinghai Lake water boundary extracted by traditional FCM clustering method,we proposed an improved RFLICM model to extract Qinghai Lake water boundary.We used the negative logarithm of conditional probability density function to describe the similarity between pixels and clusters,blurred local variables by combining with spatial position definition to control the segmentation scale of the algorithm.Then,we divided the image into lake and land by using neighborhood pixel label to define priori probability.Finally,we obtained the lake boundary through extracting the contour line of image segmentation result.The experimental results show that the improved method can effectively extract the boundary information of Qinghai Lake,with a better result compared to RFLICM and HMRF-FCM.

关 键 词：青海湖 水体边界线提取 RFLICM方法 影像聚类 

分 类 号：P208[天文地球—地图制图学与地理信息工程]