【摘要】 利用小波变换为正则化方法提供一种估计噪声能量的机制,根据估计得到的噪声能量,再利用经典正则化方法求解正则化参数,从而得到正则化解.实验结果表明带估计噪声能量机制的正则化方法,在噪声能量和原图像能量信息未知的情况下,具有较快的计算速度和较好的恢复能力.更多还原

【Abstract】 Most of traditional regularization methods need noise information, some other methods without noise have disadvantages of huge calculation or bad efficiency in some extremes. Now a new technique is proposed. In this method, we first provide an evaluation mechanism of noise energy to the algorithm by employing wavelet transform, then with the noise energy evaluation, regularization parameter and regular-ilized solution yield by using classical regularization method. Experiment results show that without noise and original figure information, the restoration effect and computation speed of our method are still good.更多还原