Research on the Super-resolution Technologies of Polarimetric SAR Images
|School||Harbin Institute of Technology|
|Course||Information and Communication Engineering|
|Keywords||Polarimetric SAR Targets decomposition Super-resolution Polarimetric Spatial Correlation|
Polarimetric SAR (PolSAR), which can provide palarimetric characteristic of the scene using four channels, i.e. HH, VV, HV and VH, is becoming more and more popular in remote sensing research area. However, due to the bandwidth of signal and dimension of antenna, resolution of PolSAR image can not compared with that of optical remote sensing image. Thus, Super-resolution processing of PolSAR images is usually desired for PolSAR image applications, such as image interpretation and target detection.Traditional super-resolution processing of PolSAR images may provide higher resolution. However, phase and polarimetric information in the original images will be discarded. Usually in a PolSAR image, each resolution contains several different scattering mechanisms. If these mechanisms can be allocated to different parts within one resolution cell, details of the images can be enhanced, which means the resolution of the images is improved. In this dissertation, a new super-resolution processing for PolSAR images is proposed, in which target decomposition and polarimetric spatial correlation are both taken into consideration. The proposed processing shows improved capability for preserving the phase information and maintaining the full polarization properties of the scatters.The processing mainly contains two steps. In the first step, different scattering components are obtained by target decomposition. The next step is to find out the distribution of different scattering components within each resolution cell by super-resolution based on quadrant pixels (SRQP) and super-resolution based on polarimetric spatial correlation (SRPSC). Generally, in remote sensing images, the adjacent pixels are more likely to represent the same scattering mechanism. Sub-pixels are considered as weighted linear combination of their quadrant pixels in SRQP. Actually, neighbouring pixels in PolSAR images have higher spatial correlation. SRPSC makes full use of the polarimetric spatial correlation to allocate different scattering mechanisms into different parts of one resolution cell. Both SRQP and SRPSC improve the resolution of PolSAR images.The proposed methods are demonstrated with DLR ESAR L-band full polarized images. The results of the super-resolution methods confirmed the effectiveness of the proposed methods.