论文部分内容阅读
针对大气湍流退化图像复原问题,提出了一种基于各向异性和非线性规整化的总变分盲复原新算法,该算法主要结合图像和湍流点扩展函数的一些性质采用基于各向异性的空间自适应规整化处理,建立了具有非线性和空间各向异性的规整化函数,使其在恢复目标图像和估计点扩展函数时能自适应地进行梯度平滑。最后,通过交替最小化方案来极小化代价函数和通过定点迭代策略将非线性方程进行线性化处理,快速地估计点扩展函数和恢复图像。在微机上对数字模拟和实际退化图像进行了一系列恢复实验,验证了算法的有效性和稳健性。
Aiming at the problem of atmospheric turbulence degenerated image restoration, a new algorithm of total variation blind restoration based on anisotropy and non-linear regularization is proposed. The algorithm combines anisotropy-based space with some properties of image and turbulent point spread function Adaptive normalization, a normalization function with nonlinearity and spatial anisotropy is established, which makes it adaptive to gradient smoothing when restoring the target image and estimating point expansion function. Finally, the crossover minimization scheme is used to minimize the cost function and to linearize the nonlinear equations by a fixed-point iterative strategy to quickly estimate the point spread function and recover the image. A series of experiments on digital simulation and degenerated image are carried out on a microcomputer to verify the effectiveness and robustness of the algorithm.