噪声条件下的运动模糊参数同步辨识的新算法

在噪声污染的情况下，匀速直线运动模糊图像频谱中的暗黑色条纹变得模糊甚至消失，根据暗黑色条纹的特征来估计运动模糊参数的方法将失效。由此，提出了一种噪声条件下的运动模糊参数同步辨识的新算法，该算法以运动模糊图像频谱作为研究对象，首先利用区域生长算法提取频谱中白色长条区域的轮廓，再计算其最小面积外接矩形，根据最小外接矩形的长度、宽度及倾斜度等参数同步估计运动模糊参数：模糊方向与模糊尺度。实验结果表明，对不同信噪比、不同模糊方向和模糊尺度的运动模糊图像，该算法可以较精确地估计出模糊参数，且具有很好的抗噪声鲁棒性。

New algorithm for motion blur parameters synchronous identification under noise conditions

Under noise conditions, the black stripes of the spectrum for the uniform linear motion blurred image become obscure or even disappear. It is difficult to estimate the motion blur parameters by means of the characteristics of the black stripes. Thus, a new algorithm was proposed for motion blur parameters synchronous identification under noise conditions, which was based on the spectrum of the motion blurred image. Firstly, the contour of the white bound area was extracted using region growing algorithm, and then the minimum bounding rectangle of the contour was calculated, the motion direction and the motion length were identified simultaneously by the length, width and tilt of the minimum bounding rectangle. The experimental results indicate that, for different signal-to-noise ratios, different motion directions and motion lengths, the algorithm can be more accurate to estimate the motion blur parameters, and has good robustness. parameters, and has good robustness.