The Passive Millimeter-wave Imaging System And The Super-resolution Of Passive Millimeter-wave Image

Millimete-wave wavelength is between microwave wavelength and infrared wavelength. Its frequency range from 30-300GHz, and corresponding wavelength range from 10-1mm. Millimete-wave wavelength is shorter than microwave,so it can get narrowband beam,and offer higher angular resolution. Compared to infared, because its longer wavelength, millimeter-wave is hardly influenced by climatic condition. So, millimeter-wave passive detector is more effective than optoelectronic detector, and it can operate under all-weather conditions, and it is becoming more and more important in the research and application area.In the process of imaging there are many factors lead to the degrade of picture, such as aberration, random atmospheric turbulence,vibration of the imaging system,blure due to target motion, etc. The aim of image resolution is to process the degrade image, and recover it. According to the viewpoint of optic Fourierism, the optic imaging system is a low-pass filter, because the effect of optic diffraction, the transfer function turns to zero when above a limit frequency which is determined by diffract limit. Obviously normal image resolution method can recover the image to the limit frequency and can not exceed it, then the frequency information beyond the limit frequency is lost.Super-Resolution is the algorithm that utilizing the method of signal processing and software to eliminate the image degrade caused by imaging system, and recover the information beyond the limit frequency, to form an image in higher definition.Projection Onto Convex Set(POCS) algorithm is one kind of Super-Resolutin, there are a closed convex constraint set to represent the speciality of the anticipant image, such as limitary energy, reliability, flatness, etc. Utilize the convex set to converge the estimate. POCS is a process of iteration, finding an estimate by POCS is equivalent to the problem of finding a point in the intersection of a number of closedconvex sets. The solution become well conditioned because it can use as much pior information as we can get. And it should be noted that the POCS is extremely flexible lies in the convex set-theoretic methods. The maximum likelihood(ML) algorithms also have many advantages,such as guaranteed convergence.Of particular significance is its good performance even in real operational environments where an accurate model for PSF of the sensing operation is not readily available due to such external factors as vibration of the imaging system,random atmospheric turbulence,blure due to target motion,etc.The ML algorithms have several advantages such as simple digital implementation and robustness of performance to inaccurate estimation of sensor parameters. However, the convergence of iterations could in some cases become rather slow and pratical implementations may require executing a large number of iterations. So, in order to improve the quality of restoration, it's better to use a priori information as much as possible. This acquires a better algorithm. In this paper we described a new algorithm based on the ML and POCS. The experiment proved its performance in image resolution.