Research On The Microwave Radiation Image Reconstruction Method Of Based On Multi-Structure And Multi-Scale

With microwave remote sensing technology is widely used in soil moisture, military target detection, radio astronomy, marine monitoring, resource development, environmental protection and many other areas, the traditional microwave radiometer is limited by low spatial resolution, it has been unable to meet the requirements of the development of microwave remote sensing fine and complex technology development. In order to solve the spatial resolution of microwave radiation image and the high cost of hardware conflicts, the compressed sensing theory is introduced, fully tapping sparse structure and the transcendental character of microwave radiation image, researching on the reconstruction method of based on multi sparse dictionary learning mixed structure. Taking into account the microwave radiation image has the partial and the whole self-similarity, fractal compression can be achieved, on the basis of sparse matrix dictionary learning proposed collage reconstruction method based on multi-scale fractal dictionary. Paper research content is as follows:1. Analysis the advantages and disadvantages of synthetic aperture microwave radiation imaging system, introduce compression perception theory model, aiming at a series of problems of traditional synthetic aperture microwave radiometer, perception of microwave radiation imaging system structure based on compression. Dig through the structural features of microwave radiation image sparse prior information, without affecting the signal reduction under the premise of fully compressed signal and reduce the signal sampling rate.Compressed sensing mainly includes signal observation, sparse representation of signals and signal reconstruction.Focuses on the sparse representation of signals and signal reconstruction.2. Complex characteristics of the microwave radiation image information is very difficult to use a single orthogonal sparse matrix dictionary for sparse representation, using the K- SVD dictionary learning algorithm, to be able to sparse structure characteristics of the image better, improve the sparse representation ability of a dictionary, the microwave radiation characteristic of complex structure image better by characterization.By difference transform and wavelet transform, the image signal transferred to the difference on the domain and wavelet domain.On this basis, put forward based on the structure of the hybrid orthogonal sparse dictionary learning method of microwave radiation image reconstruction.The experimental results show that the proposed algorithm to reconstruct the image effect is better than DLMRI algorithm and GradDLRec algorithm reconstruction quality.3. According to the microwave radiation image self-similarity structure characteristics,the introduction of the fractal image coding method.The introduction and the relatedmathematical theory of fractal geometry. In detail elaborated the fractal coding and decoding process of image reconstruction. Due to microwave radiation image with piecewise smooth and structure features of rich details, the traditional single scale fractal coding is very difficult to collage to reconstruct the high similarity of images, this paper proposes a multi-scale fractal coding collage refactoring microwave radiation image, namely no longer USES the fixed single fractal base dictionary, but construct a mixture of multi-scale fractal base dictionary,can better and R block, the optimal matching collage out higher microwave radiation image similarity.The simulation results show that the proposed multi-scale fractal image reconstruction algorithm in image reconstruction effect is significantly superior to the traditional method of fractal image reconstruction algorithm.