Research On The Method Of Millimeter-wave Imaging Super-resolution Algorithms Based On Compression Sensing

Millimeter-wave imaging system forms images by detecting the millimeter-wave radiation energy from the scene and utilizing the differences of the radiation intensity. This technology has many unique advantages. Relative to the infrared imaging, it cannot be affected by weather conditions or other factors. And compared to the microwave imaging, it has higher image resolution. So it has widespread potential applications in many areas such as scene monitoring, airport security check and battlefield reconnaissance, et al. However, the millimeter wave imaging system always has low-pass filtering effect due to the finite size of the antenna. Hence the acquired images have poor resolution. In order to improve the imaging quality, the traditional technology needs to increase the complexity of the system hardware and the cost.In recent years, a new developed theory named compressed sensing can offer a revolutionary solution. It can accurately reconstruct original sparse signal according to the part of the measurement information. The theory of compressed sensing has been introduced into the field of passive millimeter-wave imaging in this thesis, from which we can improve the performance of the imaging system and the image spatial resolution. The main contents we have researched are as follows:1. We research the method of millimeter wave imaging and the basic theory of compression perception based on compressed sensing architecture, including the basic concept of sparse representation, measurement matrix and the reconstruction algorithm of sparse optimization. All this have the guiding significance for subsequent design.2. In order to solve the problem of the slow convergence, we analyze the property of the imaging model and its performance based on block compressed sensing theory. The circulation measurement matrix with coefficient is adopted in the imaging system. And we finally do some simulation to verify its validity finally.3. For the millimeter wave imaging, we need an effect and fast algorithm for real-time imaging. To solve the problem of low resolution images caused by the low-pass effect of passive millimeter wave imaging system, we analyze the super-resolution algorithms for passive millimeter wave images named non-convex shrinking iteration algorithm for non-linear reconstruction process in this thesis. A two step iterative process has been introduced in the iterations of the algorithm. Our algorithm INCSHI has a smaller reconstruction error and faster convergence compared to NCSHI algorithm in our experimental results.4. The sparse prior information of the image is used in the super-resolution processing of millimeter wave imaging. And we research on the projected landweber(PL) super resolution algorithm for the supper-resolution process. We propose an improved algorithm named IPL to solve the problem of destruction of low-frequency component inside the pass-band. The wiener filter algorithm principle is adopted and relaxation parameter adaptive update process is introduced in the algorithm at the same time to solve the difficulty of choosing the relaxation parameter. Our algorithm is very effective when the image noise exists and has good super-resolution performance during the same time. Finally, use our algorisms to process the practical millimeter wave image and obtain significant good results.