一种改进的NAS-RIF红外图像盲复原算法

摘要

针对军事目标红外图像信噪比低、NAS-RIF算法复原模糊图像时敏感于噪声的缺陷，提出一种基于Contourlet多尺度变换去噪和图像细节规整化的改进NAS-RIF盲复原算法。首先，通过Contourlet变换对图像进行去噪预处理；然后，利用最优阈值分割技术提取目标的可靠支持域，并引入规整化方法，在代价函数中添加目标边缘保持约束项，保存图像细节特征；最后，利用共轭梯度（CG）算法优化代价函数，以保持算法的收敛速度。两组实验的结果表明，针对信噪比较低的气动红外退化图像，与原始NAS-RIF方法相比，本文提出的改进算法具有更好的复原效果，算法的收敛速度基本保持不变。%To solve the problems that infrared image of military target has low signal noise ratio and NAS-RIF algo-rithm is sensitive to noise when applied to restore degraded infrared image,a new modified NAS-RIF blind restoration algorithm based on contourlet transform denoising and image details regularization is proposed. Firstly,the image is preprocessed by contourlet transform denoising. Then,using the optimal threshold segmentation technology,the target′s support reliable domain is extracted. And target edge maintaining constrained item is added in cost function by regularization method to save detail features of the image. Finally,conjugate gradient (CG)algorithm is used to optimize the cost function,which keeps the speed of convergence. Two experiments show that:for aerodynamic infrared degraded image under a low signal noise ratio (SNR)condition,compared to the original NAS-RIF method,the pro-posed modified algorithm has better recovery effect and the speed of convergence remains almost unchanged.