基于两幅模糊与噪声图像的图像交替修复算法

An Image Alternate Restoration Algorithm Based on Two Blurred/Noisy Images

针对一幅模糊图像和一幅噪声图像的图像修复问题,提出了一种结合TV-L^1模型与TV模型的新交替修复算法。该算法首先利用TV-L^1模型对噪声图像进行去噪;然后利用TV模型,把去噪结果作为迭代初始值,对于模糊图像去模糊;最后,把去模糊结果作为迭代初始值,再利用TV-L^1模型对噪声图像进行去噪……,如此交替进行。实验效果表明,该新算法不仅继承了TV-L^1模型与TV模型能保持轮廓和细节的优点,同时也有效地克服了这两种模型会降低对比度和出现"重影"的缺点。

In this paper, the image restoration problem is studied with a given blurred image and a noisy image. We propose a new algorithm combining a TV-L1 de-noising model with a total variational deblurring model. In our algorithm, we use TV-Ll model to de-noise the noisy image in the first; Then we use the TV model to deblur the blurred image by taking the de-noised result obtained in the secod step as the initial value of the next iteration ; Finally, they use TV-L1 model again to de-nolse the noisy image by taking the de-blurred result got in the second step as the initial value of the next iteration... , and so on. Experimental results tell us that our new algorithm not only inherits the advantage of edge preserving of the TV-L^1 model and the TV model, but also overcomes its＇ disadvantage of degrading the contrast of image.