The paper make a study of small target detection in IR background and low SNR.The content of study comprise the theory of image model, morphology filter , background correlation and so on.First,the paper discusses the digital image's expression method of matrix and vector,and the common theory of image's statistic model.The research of image is imported to macroscopical statistic rule and microcosmic statistic relation from image's random field model.The paper researches measurement methods of serial images's otherness to aim at ATR problem.Infrared image backgrouding's statistic characteristic is important to image'spretreatment and target detection.The paper discusses the backgound'stationary,Gaussion and long-range correlation.To the infrared image pretreatment,the method based on morphology filter is used to wipe off disturb and yawp.And adaptive filter is used to implement signal's enhancement.Finally,the paper introduces neural network's arrangement concept and the application of BP to pattern recognition.And discusses a kind of BP network forecast based on non-linear and adjustable function.
|