旋转运动模糊图像复原的TSVD方法

TSVD Algorithm for Image Restoration Blurred by Rotational Motion

在Neumann边界条件假设下,具有对称及平移不变性的点扩散函数的图像复原问题可以转化为不适定的解卷问题,进而可以采用TSVD方法解卷。而旋转运动模糊图像的点扩散函数不具有对称性,距离旋转中心越远,图像模糊越严重,在不同的模糊路径上具有不同的模糊参数。按照半径的不同把图像分成许多同心圆,每个同心圆上的旋转模糊近似看作水平直线运动模糊,从而利用TSVD方法,进行复原,实验说明了算法的有效性。

With the assumption of Neumann boundary condition,the image restoration problem is equivalent to an ill-posed deconvolution problem with the point spread function being symmetry and translation-invariance.Then we use the TSVD algorithm to solve the convolution equation.Unfortunately,the point spread function of the rotational motion blurred images doesn＇t have symmetry.Because the images will be more blurred as the pixels are away from the center of rotation,this is to say,the pixels of images have the different blurring parameters as the pixels being from different places in the target images.In this paper,the target images are divided into many concentric circles according to the different radii,then the rotational motion blurring of each concentric circle is approximately regarded as a horizontal linear motion blurring.Then we use TSVD algorithm to restore the rotational motion blurred images,and the experimental results show that the introduced algorithm is effective and promising.