Speckle Reduction of Multiple Beam Mode RADARSAT SAR Data Using Wavelet Transform
Jung-Seon Won, Ju-Hyeong Ryu, Sang-Wan Kim · National Remote Sensing Bulletin · 1997
SAR provides excellent renditions of Earth surface, and its image quality is almost independent of weather conditions. Among several civilian SAR systems, the RADARSAT SAR has a unique capability to offer multiple beam modes, resolutions, and incidence angles. The fine beam mode of the RADARSAT SAR is specifically useful to monitor fine detailed surface features because of its high resolution. However, speckle noise in SAR image, produced by coherent illumination on rough surface, degrades quality of image and often leads to misclassification. Since it is single-look image, speckle noise in fine beam mode is normally more serious than that in standard beam mode. To reduce speckle in the fine beam mode image of RADARSAT SAR, we have developed a speckle reduction approach by adopting a weight averaging process in wavelet transform domain. The test results demonstrate how effectively the approach reduces the speckle noise in a fine beam mode image, and its performance is discussed in terms of three criteria:1) preservation of mean, 2) reduction of variance, and 3) preservation of edges. To optimize the algorithms with respect to level of wavelet transform and weight, further tests are, however, necessary.