Microwave Radiometer Spatial Resolution Enhancement Algorithm

Generally the spatial resolution of spaceborne microwave radiometer is low since its antenna size and weight are limited by its platform. Further more, in order to get geophysical parameters, the observed values from different frequency channels are utilized. So it is normally desirable to unify the different channel's spatial resolution to one higher frequency channel. Therefore it is very necessary to enhance the spatial resolution of images from radiometers by using spatial resolution enhancing methods in view of applications.In this paper, Backus-Gilbert matrix inversion technique, image deconvolution technique and scatterometer image reconstruction (SIR) algorithm are introduced to improve the spatial resolution firstly. With some synthetic images, all these methods are simulated and compared under two different conditions, without noise and with noise in a synthetic radiometer. In additon, a new method-subdivided reconstruction which improves BG inversion method to further enhance the spatial resolution is presented in this paper. The effectiveness of this new procedure has been demonstrated through various quantitative and qualitative validation tests.The effectiveness of all above methods will be degraded when there is random noise in the microwave radiometer. A denoising method using steerable pyramid representation based on Gaussian Scale Mixture (GSM) in wavelet domain and a noise estimating method using data masking are presented. Then the noise-removed images are reconstructed to improve their spatial resolution by subdivided reconstruction method.In the end, the application of spatial resolution enhancing methods in a real spaceborne microwave radiometer, which spatial sampling interval is much smaller than its spatial resolution, is analyzed.