摘要
SAR图像中的噪声是乘性的斑点噪声,对光学图像有效的检测算子并不适用于SAR图像的边缘提取。由于小波变换具有良好的时频局域化特性及多尺度分析能力,因此根据多尺度分析来构造多尺度边缘检测算子,通过改变高斯函数的标准差σ,就可以选择边缘检测的细节程度;最后,将各尺度检测结果进行融合来提高检测效果。实验结果表明,该方法取得了较好的检测效果。
The noise within SAR images is multiplicative speckle. General detecting operators which are fit for optical image can not be used in SAR image. Wavelet transform has satisfactory local characteristics and multiple-scale analysis capability for non-periodical or sudden change signals. The multi-scale detecting operator based on multi-scale analysis are constructed in this paper. We can extract different details of edge in image by changing the standard difference σ of the function of Gauss. AT last, each result is fused to improve the effection of detection. The experimental result indicates that this method has good detecting effect.
出处
《计算机应用研究》
CSCD
北大核心
2005年第7期161-162,236,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(60272022)
