运动模糊图像恢复的核心是点扩散函数的估计和直接去卷积算法。针对快速运动而形成的低信噪比和小模糊长度图像模糊的问题,提出了一种新的算法来估计模糊核函数的参数,在确定模糊核函数后,模糊图像的恢复采用了一种自然图像梯度统计先验的直接解卷积算法,实验结果证明,与R.Fergus的算法相比较,对于线性运动造成的图像模糊有更快的速度和更好的恢复效果。
The key of the blurred image blind restoration is the estimation of the point spread function and non-blind deconvolution algorithm. For the problem of the image blur which low SNR and small blur length caused by the camera shake, a fast robust algorithm is proposed to estimation the parameters of blur kernel function. Then the blurred images are restored by using a fast non-blind deconvolution method. Compared with R. Fergus' algorithm, the experiment resuits prove that the algorithm for motion blurred by the relative linear movement has faster speed and good recovery effect.