低质图像去模糊反卷积数学建模仿真

Mathematical Modeling and Simulation of low Quality Image Deblurring and Deconvolution

图像采集设备受温度偏移、光学偏移、成像时的相对运动、失焦等因素影响,可能导致图像质量下降的情况,为了提升图像视觉效果,提出基于正则化约束的低质图像去模糊数学建模方法。基于网函数插值原理和扩散理论构建迭代网函数插值算法用于椒盐噪声去除,同时联合Tetrolrt变换和主动随机场模型去除高斯噪声,提升图像质量。依据图像模糊核稀疏性和梯度稀疏性,采用混合正则化约束构建两者的估计数学模型,通过交替方向乘子法求解数学模型,并利用L1范数和全变分法反卷积低质图像,实现图像去模糊。实验结果表明,所提方法去模糊后图像峰值信噪比、结构相似性、平均梯度和视觉保真度更高,图像细节更清晰。

In order to improve the visual effect of images,a mathematical modeling method for deblurring lowquality images based on regularization constraints was put forward.Based on the principle of net function interpolation and diffusion theory,an iterative network function interpolation algorithm was constructed for removing salt-and-pep⁃per noise.At the same time,the Tetrolrt transform was combined with active random field model to remove Gaussian noise and thus to improve image quality.Based on the sparsity of image blurring kernel and gradient sparsity,a mixed regularization constraint was adopted to construct an estimation mathematical model for both.Then,the model was solved by the alternating direction multiplier method.Moreover,low-quality images were deconvoluted by using the L1 norm and total variation method.Finally,the image deblurring was achieved.Experimental results show that the proposed method achieves higher peak signal-to-noise ratio,structural similarity,average gradient,and visual fidelity of the image after deblurring,as well as clearer image details.