拉普拉斯修补矩阵下图像渐晕高效复原方法

Efficient Recovery Method of Image Halo under Laplace Repair Matrix

在图像退化模糊的过程中,噪声与干扰同时存在,导致图像复原运行慢、复原图像缺乏高频细节等问题。为解决这一问题,研究出一种基于拉普拉斯修补矩阵的渐晕复原方法。建立调和去噪模型,保留图像边缘细节和背景信息,利用贝叶斯规则计算模糊问题后验分布,交替迭代渐晕图像模糊核和中间层区域,在退化函数和加性噪声项基础上计算原始图像估计值,模型沿着等照度线完成扩散,在指定梯度上平滑图像,抑制和削弱渐晕,实现图像复原。实验证明所提方法可行有效,复原后图像清晰,噪点少,计算迭代速度快,鲁棒性好。

In the process of image degradation and blurring,noise and interference may lead to slow operation of image restoration and lack of high-frequency details of the restored image.As a result,a vignetting restoration method based on Laplacian repair matrix was proposed.At first,a harmonic de-noising model was constructed to retain the image edge details and background information.After that,the Bayesian rule was adopted to calculate the posterior distribution of the blur problem.Then,the fuzzy core and middle layer of the vignetting image were iterated alternately.The estimated value of the original image was calculated on the basis of the degradation function and additive noise terms.Moreover,the model diffuses along the isoline.Meanwhile,the image was smoothed on the specified gradient to suppress and weaken the vignetting.Finally,image restoration was achieved.Experiments show that the proposed method is feasible and effective.The restored image is clear with less noise.In addition,the iteration speed is fast,and the robustness is also good.