CAJ | 학술논문

运动模糊图像复原是模糊图像复原领域中的重要课题。首先，利用一种新的正则化方法有效估计了非参数运动模糊核。然后，在贝叶斯框架下，基于图像像素空间和梯度空间的统计特性，结合自然图像的梯度约束，提出了图像复原的改进RL正则化算法，有效抑制了复原图像中存在的振铃效应。实验结果表明，RL正则化算法在模糊核的估计存在误差时，依然能够得到较好的图像复原效果。
운동모호도상복원시모호도상복원영역중적중요과제。수선，이용일충신적정칙화방법유효고계료비삼수운동모호핵。연후，재패협사광가하，기우도상상소공간화제도공간적통계특성，결합자연도상적제도약속，제출료도상복원적개진RL정칙화산법，유효억제료복원도상중존재적진령효응。실험결과표명，RL정칙화산법재모호핵적고계존재오차시，의연능구득도교호적도상복원효과。
Motion blurred image restoration is the important subject of blurred image restoration, a novel regularization approach is introduced to estimate the non-parameter motion blur firstly in this paper. Then, based on the statistical properties of pixel space and gradient space, a modified Richardson-Lucy (RL) regularization approach for image restoration is introduced, also using the gradient constraints of the natural images. The proposed approach can suppress the ringing effect effectively in the restored image. Then experimental results show that the proposed RL regularization approach can restore the blurry image effectively, even the motion blur is contaminated by measurement errors.