摘要
介绍了INSAR干涉图的多视滤波、矢量滤波等现有的空间域滤波方法,提出了频谱加权的频率域滤波方法。该方法需要将干涉图中的干涉相位映射为单位矢量,并进行FFT变换,得到其频谱,然后采用幂函数、正弦函数或者指数函数等对频谱进行加权处理,从而抑制噪声的频率成分,最后再将加权处理后的频谱变换到空间域,计算干涉相位,得到原始干涉图的滤波结果。采用不同数据源获取的干涉图进行了矢量滤波和频谱加权滤波实验。结果表明:矢量滤波的相位保持能力较好,但是对于较差质量处和条纹密集处的干涉图滤波效果较差;频谱加权滤波属于频率域非线性滤波,它的相位保持能力较差,但是它属于全局滤波方法,能够使条纹质量较差的局部区域恢复出清晰的干涉条纹,甚至可以使没有条纹的局部区域出现清晰条纹。在滤波时要根据干涉图的实际情况和滤波精度要求选择合适的滤波器。
Some spatial filters for INSAR interferograms were introduced, such as multi-looking filter and vector filter, etc. And a new filtering algorithm by spectrum weighing was proposed. This algorithm mapped the raw intergerogram to unit vectors and transformed them to frequency field by FFT. Then the spectrum was processed with weighting function to suppress the frequency component of noise. The weighting function could be selected as power function, sine function and exponent function, etc, Finally, the processed spectrum was transformed by IFFT and the phases of filtering results were calculated. Some interferograms derived from different systems were filtered by vector filter and by spectrum weighting filter. The results show that the vector filter has the ability of preserving phases but its results for low quality interferograms are not better, and that the spectrum weighing filter can get better filtering results for low quality areas, and it can even make no fringe areas emerge clear fringes, but its ability of phase preserving is worse. We should select proper filter in terms of the facts of interferograms and the requirement of filtering precision.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2006年第9期2563-2565,2603,共4页
Journal of System Simulation
作者简介
靳国旺（1977-），男，河北滦南人，博士生，讲师，研究方向为合成孔径雷达干涉测量；
徐青（1964-），男，浙江磐安人，博士，教授，研究方向为摄影测量与遥感；
秦志远（1965-），男，河北乐亭人，博士，教授，研究方向为数字图像处理。
