摘要
为了降低运动模糊图像恢复问题的复杂度,提高计算效率,将图像恢复的二维问题转化为一维问题,可以将图像的模糊方向旋转到水平方向,这就需要识别图像的运动模糊方向。计算模糊图像任意角度的方向微分,根据微分图像灰度值的绝对值的分布特征,可以看到模糊方向和微分图像的绝对灰度值存在一定的关系,由此可以鉴别出模糊方向。在鉴别过程中,给出了精度更高的三次样条插值计算微分图像的详细方法。仿真实验表明,该方法稳定性好,具有较高的精度。
To reduce complexity of restoration of blurred images and improve computing efficiency, motion blurred direction need to be identified and adjusted to horizon by image rotation. Then image restoration can be transformed to a one- dimension problem from a two-dimension problem. First, the derivation of the image was computed in arbitrary direction. Based on distribution of the absolute value of pixels of the derivative image and the relationship between the motion blurred direction and the absolute value of pixels of the derivative image, motion blurred direction would be identified. The detailed computing processes of spline interpolation with higher precision were presented. Simulation experiment shows that the new method has good stability and high precision.
出处
《计算机应用》
CSCD
北大核心
2008年第B06期220-222,共3页
journal of Computer Applications
关键词
图像恢复
模糊方向
鉴别
灰度值
样条插值
image restoration
blurred direction
identification
grey level
spline interpolation
