基于Radon变换和Gabor变换鉴别运动模糊方向角

对于匀速直线运动的模糊图像,准确鉴别运动方向角是图像复原的关键。分析了运动模糊图像的频谱特征,介绍了Radon变换的数学原理及用其估计运动模糊方向角的思路、步骤和数值实验结果。由于实际拍摄的模糊图像在很多时候频谱特征不够明显,导致用Radon变换鉴别角度出现大的误差,为此,提出了基于Gabor变换的一种改进算法。算法运用"窗口"聚焦频谱图像中心,较好地消除了噪声干扰并克服了Radon变换的弊端,数值实验结果验证了该算法的有效性。

Identification of Motion Direction from Motion Blurred Images Based on Radon Transform and Gabor Transform

In order to recover the uniform linear motion blurred image, it＇ s necessary to identify the motion direction,Based on the spectrum characteristics of motion blurred image, the mathematic principle of Radon transform and procedures of using it to i- dentify the motion direction were introduced. Aiming at the spectrum characteristics of the real blurred images taken by digital camera is not obvious, an improved algorithm based on Gabor transform is proposed. The algorithm uses a ＂window＂ focusing on the spectrum center of the image to eliminate noise interference and overcome the drawbacks of Radon transform. Numerical exper- imental results show effectiveness of the presented method.