【摘要】 提出一种基于正则化方法的高效图像复原技术。围绕最小化正则解模糊误差,设计该技术。利用泰勒级数定性地分析怎样的正则化算子使正则解模糊误差能量较小,得出结论:通常情况下应选取低阻高通的正则化算子;利用随机理论解决正则解模糊误差能量期望值最小化问题,确定正则化参数;利用小波变换估计噪声能量,在没有噪声能量信息的情况下,新方法能进行高效的图像恢复。实验结果表明本文的恢复技术比传统方法的恢复性能好,恢复效果接近最佳且性能稳定,且不需要噪声能量信息。更多还原

【Abstract】 A new technique based on regularization method and restores image to close-to-optimal is proposed in this paper. The focus is the minimization of the blurred error of regularized image. Taylor epansion helps to choose regularization operator qualitatively, and yields a conclusion that the regularization operator should be low-stop and high-pass; stochastic theory benefits to minimize the expectation of the energy of the blurred error of regularized image, and determines the regularization parameter; wavelet transform helps to estimate the energy of the noise . Numerical results illustrate that the new method performs better than traditional methods and yields steadily close-to-optimal restoration without noise energy information.更多还原