FY-3微波成像仪10.6 GHz通道超分辨率图像重建

为改善风云三号 (FY-3) 微波成像仪10.6 GHz通道空间分辨率，提出利用超分辨率图像重建算法进行图像重建。分析并指出FY-3微波成像仪10.6 GHz通道具有过采样特征，具备重建得到高于瞬时视场图像的潜力；基于超分辨率图像重建理论，根据FY-3气象卫星轨道、姿态，微波成像仪性能参数以及工作模式等物理约束，建立微波成像仪的超分辨率成像模型，并推导计算出超分辨率成像模型参数；采用POCS算法重建得到10.6 GHz通道的超分辨率图像，采用目视比较分析、图像统计分析对重建图像进行质量评价:重建图像像元平均梯度提高26.5%，功率谱分量和提高5.7%，重建图像具有更高的空间分辨率；台风亮温分析应用显示了从重建图像可发现更微小的景物细节目标。采用超分辨率图像重建算法对FY-3微波成像仪10.6 GHz通道进行重建处理，结果有效且正确，可为数据用户提供更多可用的遥感数据，对微波遥感数据应用具有提升作用。

Super Resolution Image Reconstruction for FY 3 MWRI 10.6 GHz Band

To improve the spatial resolution of FY-3 MWRI, super-resolution reconstruction algorithm is used for the first time. Analysis on the working model of FY-3 MWRI indicates that the 10.6 GHz channel data are over-sampled, and the information included in the over-sample data is the key factor for improving the spatial resolution by super-resolution reconstruction algorithm. Based on the super-resolution reconstruction theory, the orbit, attitude and characteristics of MWRI are all used as the physics restriction, the speed of the satellite and the antenna pattern are both used to calculate the over-sample rate along the orbit, the ground sample distance dividing the flying distance in the integral time of one pixel is over-sample rate along the orbit, the speed of scanning mirror and antenna pattern are both used to calculate the over-sample rate across the orbit, and the field of view dividing the scanning angel in the integral time is over-sample rate across the orbit. Based on the over-sample rates of two directions, the super-resolution imaging matrix is estimated, the scale of the matrix is over-sample rate value along the orbit times the over-sample rate value across the orbit, and parameters of the matrix are calculated out from the metrical value of the antenna pattern by integral. The 10.6 GHz image reconstruction is conducted with POCS reconstruction algorithm. Firstly, a high-resolution image is estimated from the low-resolution image by interpolation algorithm, the new observing image is estimated by the high-resolution times the super-resolution imaging matrix, and then the process is repeated over and over based on the error, through which the high-resolution image becomes closer and closer to the factual scene. In the experiment, a real 10.6 GHz image of FY-3 MWRI is processed, the spatial resolution of the high-resolution image is estimated by artificial observation and mathematic statistic. The average grads of one pixel advances from 11.72 to 14.82, the sum of the power spectral increases 5.7%, and reconstruction image is of a higher spatial resolution. The typhoon temperature analysis shows that more details can be found in the reconstruction image, and details are similar as that in the image of 37 GHz, proving the reconstruction is effective and right. The super-resolution reconstruction is useful for FY-3 MWRI image of 10.6 GHz, which can offer more images with higher spatial resolution, accelerating the application of FY-3 MWRI.