Optical Pattern Recognition with Spatially Disjoint Signal and Scene Noise

9406922在输入信号和场景噪声空间重叠的假设下推导的Javidi匹配滤波函数和最优处理器在输入信号和场景噪声空间不相交的情况下可能表现不佳，导致系统性能很差。本工作的目的是为空间不相交信号和场景噪声设计新的最佳处理器。主要研究者将考虑在非重叠输入场景噪声中目标的一般模型，其中目标可能包含诸如未知照明、比例、旋转等畸变。由于检测器噪声或其他统计波动引起的目标上的重叠噪声将包含在该模型中。本文旨在对在空间上与目标不相交的场景噪声存在下设计的光学模式识别系统进行理论、数值和实验研究。＊＊＊

9406922 Javidi Matched filter functions and optimum processors derived under the spatially overlapping input signal and scene noise assumption may not perform well in the presence of spatially disjoint input signal and scene noise, and could lead to very poor system performance. The purpose of this work is to design new optimum processors for spatially disjoint signal and scene noise. The principal investigator will consider a general model of a target in non- overlapping input scene noise where the target may include distortions such as unknown illumination, scale, rotation, etc. Overlapping noise on the target due to detector noise or other statistical fluctuations will be included in this model. It is intended to conduct theoretical, numerical, and experimental investigation of the optical pattern recognition systems which are designed in the presence of scene noise that is spatially disjoint with the target. ***