Research Of Wide Swath And High Sensitivity Thermal Infrared Imaging Technology Based On Low-rank Approximation

Thermal infrared remote sensing technology can obtain the heat distribution of the scene,and it is widely used in environmental investigation,security monitoring and other fields.But in the case of precisely detecting of the sea surface temperature,it is not yet effective to satisfy its application due to the constraints between the performances of the imaging system.Therefore,it is necessary to solve this problem from other aspects in order to realize the large range effective detection of weak temperature difference signal in the strong background.The imaging process of infrared remote sensing can obtain a large amount of data,which usually includes a lot of noise and redundancy.In terms of data processing,lowrank matrix approximation theory aims to eliminate noise from redundant data to restore real data,which is consistent with the characteristics of infrared remote sensing data.Therefore,this project combines low-rank denoising model with array scanning imaging method to proposes a wide swath and high sensitivity thermal infrared imaging technology.Firstly,the theoretical model of low-rank matrix approximation and its applicable conditions in denoising field are analyzed.Based on this,the imaging scheme and the complete data processing flow are designed.Then develope the imaging system according to the scheme and study the relevant control methods.Finally,the performance test analysis and remote sensing experimental verification are carried out.The results of the research are as follows:1.Aiming at the effective acquisition of thermal infrared image information in strong background and weak signal scene,we put forwarded a method of combining low-rank denoising model with array scanning imaging.Under the condition of no additional detector cost and cold optical design,this method obtained high denoising performance by using high redundancy data as the pointcut.2.In order to widen imaging width,the wingspan scanning direction was set as the direction of ordered motion.By using the array scanning imaging technique,the time series obtained by different pixels in the direction of the wingspan was used as the column vector of the observation matrix.Finally,multi-frame and wide swath images in the same scene were reconstructed through low-rank denoising model.3.In the actual remote sensing process,there is a sub-pixel overall shift between multiple frames in one scanned line.The image quality was improved on the original basis by using this shift to reconstruct the super-resolution image of the sequence image,which made up the shortage of spatial resolution reduction after low-rank denoising.4.A complete set of wide swath and high sensitivity thermal infrared imaging system was designed and implemented.This system achieve stable whiskbroom imaging while isolating flight attitude disturbance.It can also compensate for the forward motion adaptively.Then,the 55 degree imaging FOV and system NETD lower than 10 mK were obtained.Combined with actual remote sensing experiment,the feasibility of the wide swath and high sensitivity thermal infrared imaging detection was verified.