Research On Infrared Focal Plane Array Detector Key Technology

As the rapid development of science technology, IRFPA(infrared focal plane array) detector is coming into being.The existence of IRFPA detector is an epoch-making revolution in infrared imaging history.Meanwhile,infrared imaging system is widely used in civil fields and military fields because of its appearance.In practical applications,because processing materials and making art have an effect on detector,nonuniformity and bad pixel between detector units arise and thus it leads to the fall of imaging quality,restricts the applications of IRFPA detector.Therefore,nonuniformity and bad pixel must be resolved on imaging systems of infrared focal plane array nowadays. The history,present condition and development trend are elaborated detailedly.The paper analyzes the producing reasons,characteristics of nonuniformity and bad pixel.The paper discusses a few common methods of nonuniformity,bad pixel detection and bad pixel compensation,and indicates their merit,shortcoming and their applying situation.Aimed at the problems,the paper engages in research.The contents of research are mainly as follows:1. The problem of nonuniformity exists imaging systems which IRFPA detector is dominated, based on the advantages and disadvantages of existing nonuniformity correction algorithms are analyzed in detail,the paper draws an conclusion that neural network nonuniformity correction algorithm is the main direction in the future.Consequently,the paper adopts neural network to correct the nonuniformity of IRFPA detector.The adoption of the method has following reasons:Firstly,the method needn't demarcate IRFPA detector.Secondly,the requirements of linearity and stability of dectctor parameters don't have enough.Finally,the more importance is that correction parameters are continuous updated by neural network.2. During the neural network nonuniformity correction is used and when we look on four-closer fields' average of all image units which belong to nonuniformity image as expected output, there is a stubborn problem that image becomes fuzzy and don't maintain the properties of border. Amied at the problem and combined nonlinear middle filter with the best robustness and the advantage of image border maintenance which can fit all kinds of noise models,the paper presents an improved nonuniformity correction method.The new method acts middle value of image units set in a window...