一种基于小波变换的雷达图像边缘提取方法

图像最基本的特征是边缘,边缘特征提取是多传感器信息融合与景象匹配中的重要内容.合成孔径雷达图像通常带有较强的相干斑噪声,用传统的边缘检测算法效果很不理想.本文针对合成孔径雷达自身的特点,利用小波变换的特性,提出一种将小波变换的多尺度分析与模糊加权中值滤波相结合的边缘特征提取方法.实验证明方法有效,边缘定位准确,并对噪声有抑制作用,将提取结果用于匹配,定位精度达到1个象素.

A SAR Image Edge Extraction Method Based on the Wavelet Transform

Edge extraction is one of the most important components of multisensor information fusion.For SAR images the performance of conventional edge extraction methods such as those based on zero crossings of the second order derivatives is terribly bad due to the existence of speckle.Based on the analysis of the characteristics of synthetic aperture radar(SAR)images,this paper puts forward a combined method for edge extraction of SAR images,which is composed of wavelet transform and fuzzy weighted median filter.Wavelet can be used for edge detection because multiresolution wavelet decomposition of an image yields detailed coefficients that contain the high frequency content of the image.Median filter is a kind of nonlinear filters that has good ability in keeping edge.The experimental results show the effectiveness and stability of the method.The examples of SAR image edges and visible light image edges are given and analyzed respectively.Furthermore,these edges are used for image matching.The precision of location can come to 1 pixel.