Research On Image Non-uniformity Correction And Optimization Based On Thermal Imaging Technology

The principle of infrared focal plane array(FPA)imaging is the passive reception of thermal radiation emitted by the target,and imaging is performed based on the temperature distribution characteristics of the target.Compared with visible light imaging,it has the characteristics of strong targeting,wide detection range,and strong ability to identify disguised targets.However,due to the limitations of manufacturing processes and material levels,the response characteristics of each detector unit in the FPA exhibit differences,leading to non-uniformity and blind elements,which severely affect the imaging quality of infrared images and interfere with subsequent target detection and recognition work.Therefore,research on the correction and optimization of non-uniformity issues in infrared thermal imaging is of great significance.This paper analyzes the factors that affect the imaging quality of infrared thermal imaging systems and conducts algorithmic research on various factors.The quality of imaging is optimized through image preprocessing algorithms such as dynamic range compression,blind element compensation,and contrast enhancement algorithms.In view of the non-uniformity issue,this paper proposes a neural network-based nonlinear filter residual estimation and correction algorithm by studying the mechanism that produces non-uniformity.This method first uses a nonlinear filtering method to filter out the non-uniformity of a single column image and calculate the actual residual with the original image.Then,the current residual is obtained using the predicted residual of the previous frame and the actual residual.Finally,local dynamic scene change parameters are introduced to control the adaptive correction calculation of gain and offset coefficients.Experimental results show that this method effectively suppresses false image problems and non-uniformity.In addition,in response to the proposed solution,the elements of infrared thermal imaging image enhancement algorithms and non-uniformity correction algorithms were extracted,and an upper computer software and image processing algorithm dynamic link library for infrared imaging systems were designed to construct an infrared thermal imaging experimental platform.By collecting real scene infrared images using the experimental platform,the proposed algorithms were comprehensively analyzed with commonly used typical algorithms.The experimental results show that in general scenes,the algorithm proposed in this paper filtered out 61.3% of non-uniformity in 200 test images,and the peak signal-to-noise ratio(PSNR)of the image obtained by testing the algorithm in a uniform radiation field was 40.7,which was 13.4% higher than that of the original image.In comparison,the non-uniformity correction effect of the algorithm proposed in this paper is more obvious,and the corrected results are closer to the ideal image.