一种噪声无关的图像复原算法研究

RESEARCH ON A NOISE-INDEPENDENT IMAGE RESTORATION ALGORITHM

目前获取数字图像的方法存在使图像质量明显下降的问题,迫切需要图像复原的方法,对图像进行重建。在已知点扩散函数的情况下,通过摒弃传统的依赖噪声求出增益矩阵的过程,从修复增益矩阵出发,使复原后的图像更接近原始图像。新方法在递推过程中矫正增益滤波偏大或偏小的现象,从而减少因为缺少噪声信息而产生的误差,而且相较于其他迭代算法,该方法时间复杂度较低。实验结果表明新方法可以有效地保证复原图像质量,并提高效率。

In order to solve the problem of manifest degradation in image quality while accessing to digital images,it is urgently need an image restoration technique to reconstruct the image. Under the condition of the point spread function is known,we get rid of the traditional progress of calculating the gain matrix relying on noise,through this and proceeding from restoring the gain matrix,we make the restored image more approaching the original image. The new method rectifies in recursion process the phenomenon of either larger or smaller in gain filtering so that the errors generated caused by lack of noise information are reduced. Moreover,the new algorithm has lower time complex than other iterative algorithms. Experimental result shows that the new method can effectively guarantee the quality of the restored image and raises the efficiency as well.