Single Image Based Nonuniformity Correction Method in Infrared Camera

In infrared imaging system,the low-frequency nonuniformity noise seriously affects the imaging performance of infrared system and traditional calibration-based methods can’t effectively remove it.A low-frequency nonuniformity correction method based on surface fitting model is proposed in this paper.Firstly,noisy image is filtered by wavelet to suppress high frequency information.Then,using the sparse characteristics of gradient information of scene and low-frequency nonuniformity noise,l1 regularized energy functional for surface parameters of nonuniformity noise is established.The ADMM method is used to solve the optimal nonuniformity noise surface parameters.Finally,the estimated low-frequency nonuniformity noise is subtracted from the original image to obtain the corrected image.In this paper,the scene with simulated image and real image are corrected respectively.Infrared images are obtained by the middle wave infrared thermal imager to correct simulated images and actual images.The experimental results show that the algorithm can significantly reduce the roughness and nonuniformity of images and effectively remove the low-frequency nonuniformity noise.