Assessment Of Sampled Imaging System Based On Information Theory And The Research On The System Technologies

Research targets of this thesis include information-theory based assessmentof sampled imaging system, integrated optical and electrical system designs and processing methods for aliasing noise.The theoretical basis for introducing information-theory into sampled imaging system is discussed. Information rate measurement for an optical system is discussed. Information rate of image can be calculated by information-theory when intensity distribution of an image is connected to its probability distribution. Sampled imaging system is consist of optical systems and electrical systems, in which information rate is a common parameter for both optical and electrical systems. Information rate is also an ideal parameter for integrated optical and electrical system designs.The mathematical model for sampled imaging system is derived, and then, the application analysis model is developed focusing to practical environment. System assessment is done by computer simulation. Parameter of assessment is end-to-end average mutual information. According to information-theory, the optimized designs are the system designs which guarantee the maximum mutual information. Relative to parameter of fidelity, mutual information is more sensitive to the variance of image quality. The amount of mutual information is closely connected with sharpness, clearness and the edge effect of an image. System designs can be evaluated and optimized according to mutual information because they are function of system design parameters.Integrated optical and electrical system designs based on information-theory is one of the research targets in this thesis. Modem sampled imaging systems usually suffer from undersampling. The tradeoff between aliasing noise and optical blur determines the quality of image. General matching features and matching conditions between optical imaging system and CCD detection systemare proposed, which can be taken as the theoretical basis of integrated optical and electrical system design. The relationship of sampling interval with target statistical characteristics is analyzed. The research results show that the system end-to-end mutual information reaches its peak value when sampling interval is near to the mean spatial details of the target, where, the image quality is best.The small-scale artifacts generated by aliasing noise exist everywhere in an image, which bring difficulties to some applications, such as target identification. Aliasing effects are specially heavy in the specific application environment of space remote sensing imaging. Eliminating aliasing noise algorithm used in push-broom CCD image gathering system is derived and optimized. Strategies for CCD push-broom scanning and methods for data processing are proposed. Both of them are used to obtain several images that have different displacements along two directions although CCD array can only move in one direction in space imaging system. Aliasings in image are eliminated.Finally, rich experimental results of imaging processing verify the matching conditions between optical system and electrical system. And another group of experimental results verify processing methods for eliminating aliasing noise.