Research On The Synthetic Aperture Microwave Radiation Imaging Method Of Sparse Interferometry

The Synthetic Aperture Imaging Radiometers (SAIRs) is to sample visibility functionbased on the Nyquist theory, which does not need the mechanical scanning and can directlyimage. Due to the low imaging resolution and complex system structure, the SAIRs practicalapplication is limited seriously. According to the characteristics of image is sparse or can besparse representation in transform domain, the Compressed Sensing (CS) theory usesmeasurement matrix to extract the signal randomly, so the quantity of the projectionmeasurement data is far less than that by Nyquist sampling method. Considering the situationsof microwave radiation interferometry conducted in the frequency domain, has thecharacteristics of low frequency information richer and high frequency information less, andthe distribution of low frequency and high frequency is centralized, this paper puts forwardthe variable density sparse interferometry method based on Compressed Sensing, and thenresearch the microwave radiation image reconstruction algorithm. The main work includes:1) The experienced variable density interferometry method. In the method, the image’sfrequency domain information is blocked, and the variable density scheme is used to samplingbased on the principle of different blocks possess different information. By this way, thesampling resource is allocated to the important frequency region in the case of the totalsampling rate is same, and the interference measurement result is optimized. However, thisapproach is always used in the images that the spectrums distribution are relativelyconcentrated.2) Study of the microwave radiation image sparse interference measurement model. Dueto that the microwave radiation images have the characteristics of complex data structure andrich texture information, it is difficult to realize the super-sparse interferometry directly. Sothis paper studies the adaptive super-sparse interferometry method of one dimensionalmicrowave radiation signal firstly. According to the incoherence constraint betweenmeasurement matrix and sparse basis, we establish the incoherent optimization model whichcould adaptively obtain the sampling point probability, overcome the disadvantages of equalprobability random sampling methods, and realize the purpose of super-sparse interferometry.On the basis, the tensor product dimension reduction method is used to map the highdimensional spatial data to lower-dimensional space, achieve the two-dimensional microwaveradiation image super-sparse interferometry, and obtain the optimal interferometry results.This way is equal to optimize the number of sparse antenna array, breakthrough the limitationof actual imaging system, further reduce the amount of data collection. 3) Research on the microwave radiation image reconstruction method. According to thegradient sparsity and the local smoothness of microwave radiation image, we establish theimaging model based on total variation regularization constraint, and use the alternatingiterative algorithm to realize the optimal microwave radiation image reconstruction.The simulation and experiment results show that without increasing the complexity ofhardware, it is accurate to reconstruct high resolution microwave radiation image combinedwith the two-dimensional super-sparse interferometry method and the alternating iterativealgorithm, and it is very effective when the sampling rate is very low.