Scene-based Infrared Image Research On Improved Algorithm For Non-uniformity Correction

With the continuous development of infrared thermal imaging technology,infrared thermal imaging systems have been widely used in various industries.However,due to the influence of the process,the existing infrared system will introduce a lot of non-uniform noise,which makes the visual effect of the infrared image worse,and also restricts the later application of the infrared image.At present,there are various methods for removing non-uniform noise of infrared images,which can be divided into two categories: calibration method and scene method.The scene method adjusts the algorithm correction parameters in real time according to the scene information captured by the infrared detector.It can ensure the continuity of infrared system work,so it has gradually become the development direction of infrared image non-uniformity removal algorithm.However,infrared images processed by scene-based correction algorithms are prone to ghosting and image degradation.In this paper,after analyzing and simulating the existing scene correction algorithm,four improvements are proposed.First,an improved filter based on guided filtering for the desired image acquisition is proposed,which makes the desired image noise smaller while protecting the edge;After the existing algorithms are prone to ghosting and image degradation,a new adaptive step size calculation method is proposed.The algorithm uses different parameter iteration steps in different image areas;then,the existing algorithms often require a long time Only by performing parameter iterations can a satisfactory correction effect be obtained.To shorten the convergence time of the algorithm,a correction idea combining baffle and neural network is proposed to improve the convergence speed of the algorithm.Finally,after simulating the above improved algorithms,this paper proposes an improved weighted algorithm with better performance,and uses this weighted algorithm as the final algorithm in this paper.Through a separate simulation experiment on the final algorithm in this paper,the algorithm in this paper can reduce the non-uniformity of the infrared image,improve the imaging quality of the infrared system,and verify the effectiveness of the algorithm in this paper;subsequently,through the comparison experiment between the algorithm of this paper and other existing algorithms,It is found that the algorithm in this paper has certain advantages in both subjective vision and objective indicators.In the infrared scene of the three 1500 frame images used in the comparative experiment in this paper,the infrared image processed by the Our-NUC algorithm in this paper is compared with the Har-NUC algorithm.Compared with the Fan-NUC algorithm and the Ron-NUC algorithm,the image smoothness index ? decreased by 14.98%,17.79%,and 27.29% respectively,and the root mean square error index RMSE decreased by 14.86%,19.02%,and 29.36%,respectively.The Our-NUC algorithm in this paper converges quickly at the beginning of the algorithm,which has obvious advantages compared with other comparison algorithms,which shows the superiority of the algorithm in this paper.