摘要
提出了一种基于小波与数学形态学提取道路信息的方法。首先,利用小波多尺度分析对遥感图像进行分解,通过分析小波系数,对各个小波分量进行滤波处理,滤除非道路特征,再进行小波重构,然后对重构图像进行分割,得到包含道路信息的图像。最后应用数学形态学方法,选取合适的结构元素,对图像分割的结果进行形态变换,进一步滤除了非道路信息,完成道路信息提取。通过实验比较证明,该方法比一些典型边缘提取算法更有效、更适合于道路信息的提取。
A wavelet transform and mathematical morphology based method for roads extraction is introduced.It characterizes the roads＇ feature in the remotely sensed image through analyzing the wavelet coefficients in the multi-scale wavelet transforms.Then the image is segmented into the roads＇ region and the other easily,follows by the selection of roads＇ wavelet coefficients and wavelet reconstruction of the image.But the segmentation is not always ideal,therefore morphologic transformation is used to purify the roads＇ information.Experiments prove that this method can extract roads information more effectively and suitably than the conventional edge detected algorithms.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第22期241-243,248,共4页
Computer Engineering and Applications
基金
上海市科委重点课题资助(No.05DZ12006)。
关键词
遥感
特征
图像解译
道路提取
小波
数学形态学
remote sensing
feature
image interpretation
road extraction
wavelet
mathematical morphology
