Research On The Key Technology Of Infrared Point Target Quantitative Detection

Point target detection has always been a popular subject in various fields of military and civil satellite remote sensing.Enhancing the spaceborne infrared system's detectability of point target plays a very importance role in our country's space engineering.It leaves considerable room to promote the detectability of point target by optimizing sensor's figure of merit and data processing algorithms on account of the particularity of this kind of detection.The quantitative level of detection is especially low for the moment.Aiming at addressing the insufficiency,this paper elaborates the effects of target detection performance under different sensor channels and sampling systems by imaging simulation model and method,suggests a modified point target detection algorithm with single frame and an inversion algorithm of point target's sub-pixel characteristics in the absence of prior information.The shackles of the conventional detection research are broken.The methods of elevating the quantitative level of point target detection are discussed from different perspectives and the quantitative detection of sub-pixel target is studied.The related research achievements can provide reference for the design of spaceborne target detection systems and data processing solutions.The main results and innovations of this paper are listed as follows.(1)In the aspect of channel selection of spaceborne target detection system,infrared radiation characteristics of target and earth-atmosphere background with or without clouds are analyzed and computed in detail.A channel optimization method is designed based on optimal SCR.Some conclusions of optimal channels on top of atmosphere are drawn with regard to the early detection of target surrounded by different clutter conditions.(2)In the aspect of sampling system analysis,a sampling model of sub-pixel target is produced.It remedies the existing research methods in numerical simulation.The advantages and disadvantages of the performance of over-sampling of two times IFOV are studied and verified from many aspects: over-sampling of two times IFOV suppresses target signal and SCR,so it is not suitable for searching target under low SCR;it has to be noticed that over-sampling collects apparently superfluous radiation intensity.(3)In the aspect of point target detection algorithm with single frame,a concept of valid pixel detection is explicitly proposed.A SCR criterion based on complex background separability measurement is presented to distinguish target.The performance of the algorithm in structured background with step characteristics is better.Besides,the label field probability response model of MRF to Euclidean space distance was built,and the response ability of the target probability to neighborhood label change is improved by a higher order energy function.The target's radiation-dimension detection ability of the modified algorithm is more vigorous.(4)In the aspect of target's characteristics inversion,the target detection problem is promoted from qualitative level to quantitative level.An inversion algorithm of target's sub-pixel characteristics in the absence of prior information is proposed under certain assumptions.It is verified that the inversion algorithm can accurately estimate the target temperature and area ratio.