水下数字图像盲复原算法研究

图像复原的目的是从观测到的退化图像重建原始图像，它是图像处理、模式识别、机器视觉等的基础。盲复原作为其中一个重要分支，其主要思想是在点扩展函数未知的情况下，力求获得最佳的清晰效果。由于水下图像退化模型中点扩展函数一般为高斯模型，故针对此提出误差一参数估计法，根据图像退化过程，给出频率域的误差形式，并选定参数的变化范围，再利用复原算法做出误差参数曲线，由此估计出点扩展函数的参数值，最后利用经典的复原算法，如维纳滤波对退化图像进行复原。实验证明该方法获得了比较清楚的复原效果。

Research on Blind Restoration Algorithms of Underwater Image

The goal of image restoration is to rebilud primitive image from degenerated image which is observed,it is the foundation of image processing, pattern recognition, machine vision and so on. As an important branch, the main thought of blind restoration is to acquisitive best clear effect in situation of unknown of the point spread funtion. As usual, the PSF is Gaussian model in underwater image degeneration model, so error - parameter estimation is made, it＇s according to the degenerative process of the image, provides the error form of the frequency land and changes the range of selecting parameter, and then uses the restoration algorithms to make a curve of error - parameter,and estimates parameter values, then uses classic restoration algorithms as Wiener filter to restorate image. The experiment proves this method has obtained clearer result.