Statistical Characteristics In Synthetic Aperture Microwave Radiometric Detection

Synthetic aperture microwave radiometers can be applied for detecting aerial targets. They hold many valuable features, including concealment and survival ability, day and night vision, weather independence, and battlefield applicability. High-spatial-resolution and instantaneous imaging without scanning is achieved by adopting the technique of aperture synthesis. So the researchers pay more and more attention to the novel detection method. To evaluation the detection performance of this method, the study of statistical characteristics is carried out on three aspects: microwave radiation signals, visibility measurements, and inversed images of synthetic aperture radiometers. Main works are given below:The statistical characteristics of microwave radiation signals are analyzed. A microwave radiation brightness temperature map model is presented, which simulates the target, the sky, and the earth. To account for the effects of the atmospheric refraction and attenuation and the earth curvature, a sectionalized microwave propagation algorithm based on exponential-layered spherical atmosphere is presented. The microwave radiation characteristics of the target and the background at low elevation angle can be analyzed quantitatively by this model.Beginning with the statistical description for the predetection signals of the receivers, a statistical model of visibilities in digital synthetic aperture radiometry is derived from the compex Wishart distribution. The probability density function of visibilities is expressed in closed form, considering the impacts of the predetection filter shape, the sampling frequency, and the quantization resolution. It has been shown that compensation with an effective sample size can correct the model error induced by the nonnormality and statistical dependence of signal samples. A method of amplitude moments is proposed to determinate a suitable value for the effective sample size.After analyzing the ill-posedness of the image inversed from the visibilities, a unified regularization framework for inversion algorithms is established. Two improved linear discrete regularized inversion methods are presented respectively: the adaptive Tikhonov regularization adaptive Tikhonov regularization method, introducing the L-curve criterion to select optimal parameter adaptively, and the damped singular value decomposition method, utilizing a new smooth filter factor to suppress excessive ripple. Compared with the traditional FFT inversion, both two methods can reduce noise and distortion of images.Based on the distribution of visibilities, the statistical distribution model of inversed images is proposed further. The explicit probability density function of the pixel values is given, which is not only available for the FFT inversion, but also for all linear inversion algorithms. The signal noise ratio of inversion methods can be evaluated by this model.The statistical model of inversed images is applied for investigating the performance of target detection. The detection probabilities and the false-alarm probabilities of three detection approaches are discussed, including single-frame detection method, and detect-before-track and track-before-detect multi-frame algorithms. Taking into account the microwave radiation characteristics of the target and the background, the detection distance equation is presented, and the detection distances of three methods are compared with each other. The research in this paper is much of help in performance evaluation and parameter design of synthetic aperture microwave radiometric detection systems.