基于小波变换的SAR图像边缘提取新方法研究

由于SAR图像含有相干斑噪声，使得常规方法应用于SAR图像边缘检测时遇到了较大的困难，提出了一种新的SAR图像边缘提取方法，该方法首先用基于结构信息的统计滤波方法对SAR图像进行滤波，然后利用小波变换具有多分辨分析的特性，并结合模糊中值滤波、阈值化处理及微分梯度算子对SAR图像进行边缘提取。实验结果表明了该方法是一种有效的对SAR图像进行边缘提取的方法。

On Good-Quality Edge Detection of SAR Image

As SAR image contains speckle noise, traditional methods of edge detection find it difficult to obtain good edge detection of SAR image. We propose a new method of SAR image edge detection based on wavelet transform. Section 1 discusses the shortcomings of existing filters and then proposes a statistical filter based on structure information . Fig.1 shows the results of different filterings; Fig.1(b) for Lee filter, Fig.1(c) for Sigma filter, Fig.1(d) for Gamma MAP filter, and Fig.1(e) for statistical filter based on structure information. As compared with the other three filters, statistical filter based on structure information can reduce noise of image and get better edge detection of image. According to Refs.4 and 5, wavelet decomposition of image can divide it into one low frequency image and three high frequency images; according to Refs.4 and 5, such decomposition is beneficial to edge detection of image. We apply wavelet theory combined with fuzzy middle value filter and differential operator to the extraction of edge in SAR image. Fig.2 gives the results of edge detection of SAR image obtained with four different methods: Fig.2(b) for our new method,Fig.2(c) for traditional Canny operator, Fig.2(d) for traditional Prewitt operator, and Fig.3(e) for traditional Roberts operator. Fig.2 shows preliminarily that our new method is an efficient way for getting good quality SAR image edge detection.