论文部分内容阅读
不同波段的极化合成孔径雷达(SAR)图像间的配准,是多波段极化SAR数据融合中的一个重要问题。该文从地物的极化散射机理出发,提出了一种适用于尺度不变特征变换(SIFT)配准算法的极化特征。该特征包含了地物目标主要散射成分的信息,并反映了其他弱散射成分的强度分布,可在不同波段极化SAR图像中保持稳定。实验结果表明:与使用散射总功率(Span)实现多波段极化SAR图像配准的方法相比,该特征在不同波段下的差异较小;使用SIFT算法配准后,该特征图像可得到更多的关键点和正确配准点,且配准点的分布较分散,从而有效地提高了多波段极化SAR图像的配准性能。
The registration of Polarimetric SAR images in different bands is an important issue in multi-band polarization SAR data fusion. Based on the polarimetric scattering mechanism of ground objects, this paper presents a polarization feature that is suitable for the SIFT registration algorithm. This feature contains the information of the main scattered components of the object and reflects the intensity distribution of other weakly scattered components, which can be kept stable in different bands of polarized SAR images. The experimental results show that the difference of this feature in different bands is small compared with the method of using Span to achieve multi-band polarimetric SAR image registration. After the registration with the SIFT algorithm, the feature image can be more Many key points and correct registration points, and the distribution of registration points is relatively scattered, which effectively improves the registration performance of multi-band polarimetric SAR images.