The application of discrete wavelet transforms to SAR image processing

• Main Author: Zeng, Zhaohui
• Language: English
• Published: 2009
• Online Access: http://hdl.handle.net/2429/10152

Synthetic aperture radar (SAR) is a very efficient instrument for obtaining a. better understanding of the earth's environment. SAR, data represents an important source of information for a large variety of scientists around the world. However, the acquiring mechanism of SAR is quite different from other sensors, such as optical sensors. It brings some unique properties of SAR image data which decides that conventional image processing technique may fail to obtain satisfactory result or have to be modified to adapt the application of SAR image data. The objective of this thesis work is to investigate the potential of discrete wavelet transforms (DWT) for SAR image processing. The emphasis is placed on speckle noise reduction and SAR image compression, which are the two of the most popular application fields of DWT to image processing in the current literature. Two new algorithms for speckle reduction have been developed: a Bayesian method based on the statistical model and wavelet extrema based on the local property of wavelet coefficients, and have been applied to both airborne and spaceborne SAR images. The comparison of their results to some existing well known methods show their advantages on both the visual and numerical sides. In addition, simultaneous speckle reduction and data compression can significantly improve the compressibility of SAR images. The modified SPIHT aJgorithm has been applied to SAR image coding. The effectiveness of this strategy has been proven from the comparison to the method without speckle reduction and classic efficient wavelet compression algorithms.