摘要
研究了如何有效抑制合成孔径雷达图像中的斑点噪声并尽量保留图像的细节信息。选用具有较好紧支性和正交性的Daubechies小波,对图像进行2D小波变换,并在对各层小波系数阈值处理时采用了软硬阈值方法以及改进的软硬阈值折衷方法。实验表明,小波阈值去噪较许多常用的斑点去噪方法具有很大优势,而软硬阈值折衷方法可以一定程度上结合软、硬阈值处理的优点,既能较好地滤除SAR图像中的斑点噪声,又具有较好的边缘保持能力,从而改善了图像的解译能力。
The method of speckle noise reduction in the Synthetic Aperture Radar （SAR） images and retaining useful detail information was studied. Daubechies wavelet which has good compact support and orthogonality was chosen to make SAR images 2D wavelet transformation. Three schemes were used to shrink the wavelet coefficients, namely hard-thresholding, soft-thresholding, and a tradeoff between soft-thresholding and hard-thresholding. The result showed that the wavelet threshold has many advantages, compared with many popular algorithms for speckle reduction. The tradeoff between soft-thresholding and hard-thresholding can not only remove the speckle noise well, but also have a good edge saving capacity, and then the interpretation of the images can be improved.
出处
《装备环境工程》
CAS
2008年第1期61-65,共5页
Equipment Environmental Engineering
