基于高低阈值Canny算子的遥感影像水域边缘检测方法  

Remote sensing image water body edge detection method based on high-low threshold canny operator

作　　者：张健 ZHANG Jian(Zhejiang Brigade of China Building Materials Industry Geological Exploration Center,Hangzhou,Zhejiang 310000,China)

机构地区：[1]中国建筑材料工业地质勘查中心浙江总队,浙江杭州310000

出　　处：《北京测绘》2025年第4期496-502,共7页Beijing Surveying and Mapping

摘　　要：为保证在处理各种复杂的遥感影像中,检测过程能对不同的遥感影像的特点和需求做出自适应调整,提高边缘检测精度,本文提出基于高低阈值坎尼(Canny)算子的遥感影像水域边缘检测方法。在传统的Canny算子中采用数学形态学滤波进行遥感影像平滑去噪后,进行遥感图像梯度运算时增加对倾斜方向梯度信息的统计,依据遥感影像梯度运算结果,采用改进的大津(Otsu)法,自适应选取高低阈值筛选水域边缘点;再进行水域局部边缘细化,遍历整个遥感影像水域,滤除伪边缘,并连接真实边缘线,实现高质量水域边缘检测。实验结果说明,该方法可检测出不同的水域遥感影像边缘;且去除噪声的性能优于其他方法,失真程度更小。To ensure that the detection process can adapt to the characteristics and requirements of various complex remote sensing images,improving edge detection accuracy,this paper proposed a water body edge detection method for remote sensing images based on the high-low threshold Canny operator.In the traditional Canny operator,after applying mathematical morphology filtering for image smoothing and denoising,the gradient computation of the remote sensing image was enhanced by adding statistical information on the gradient in the tilted direction.Based on the gradient computation results,the improved Otsu method was used to adaptively select high and low thresholds for filtering water body edge points.The method then performed local edge refinement in the water body,traversed the entire remote sensing image water body,removed false edges,and connected real edge lines to achieve high-quality water body edge detection.Experimental results show that this method can detect edges in different water body remote sensing images,and its noise removal performance is superior to other methods,with minimal distortion.

关 键 词：高低阈值 坎尼(Canny)算子 遥感影像 水域边缘检测 数学形态学滤波 梯度计算 边缘检测 自适应 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] P237[自动化与计算机技术—计算机科学与技术]