On Good-Quality Edge Detection of SAR Image

xue xiaorong,zhang yanning,zhao rongchun,chen yi

DOI: https://doi.org/10.3969/j.issn.1000-2758.2003.03.021

2003-01-01

Abstract: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.