Research On Low Noise Infrared Imaging And Super-resolution Reconstruction

Due to the rapid progress of infrared imaging technology,it has been widely used in monitoring,medicine and remote sensing.Due to the factors of the hardware system and the application environment,the acquired infrared image often suffers from blurred edges,high noise,and low resolution.For the purpose of reducing the noise of infrared imaging circuit and correcting the non-uniformity of the images,the overall design of the circuit based on FPGA was performed in this article.The circuit noise was reduced by setting the bias voltage of the imaging circuit,temperature control of the infrared focal plane detector,precise design of the circuit board,and shortening of the lead wire,and the non-uniformity of the image was corrected by comprehensively applying a least mean squares(LMS)adaptive algorithm and a two-point temperature calibration method.Through the hardware system debugging,the imaging results were good.The non-uniformity correction algorithm simulation results showed that the average residual non-uniformity of the infrared image was reduced by 21% to 25%,and the PSNR was improved relatively by 3.3dB to 3.9dB,by measuring various parameters of the detector,it was proved that the imaging system meets the design requirements,noise equivalent temperature difference(NETD)was 55 mK.In order to obtain high-resolution infrared images,an improved algorithm based on the POCS algorithm was proposed through the study of super-resolution image reconstruction algorithms.In the process of image reconstruction,each low-resolution image was processed by least squares method and Laplace enhancement method firstly.The noise was preprocessed by least squares method,and the edge details of the sequence images were enhanced by Laplace enhancement method in advance.In the process of image registration,control parameter was added to control the convergence of the algorithm and thresholds was added to improve the stability of the registration.In the process of image reference frame correction,the weighted PSF was added to correct the edge pixels of the image in different degrees.Through simulation experiments,it was proved that the accuracy of the improved algorithm is within 0.1 pixels,and the peak signal-to-noise ratio(PSNR)of the reconstructed image was greatly improved by 6d B to 10 dB and the operating speed was increased by 2s compared with traditional algorithm,by reconstructing the measured sequence low-resolution images,the reconstructed images were clear,and the PSNR was relatively improved by 5dB to 6dB and the operating speed was relatively increased by 2s to 3s.