ZHU Li-fu, YAN Bi-xi, WANG Jun, et al. Restoration technique of blurred image caused by jitter of visual sensor[J]. 2018, (8): 19-21. DOI: 10.13873/j.1000-9787(2018)08-0019-03.
视觉传感器抖动模糊图像复原技术
摘要
针对视觉传感器的抖动导致采集图像模糊不清的问题
研究了现有的模糊长度和模糊方向参数鉴别的方法。根据运动模糊图像背景像素点的相关性
对模糊图像采用Sobel算子进行一阶微分
并进行自相关运算
相关曲线每一列的两个边界值求和
求和曲线中两个相等的最小值的距离等于运动模糊长度;模糊图像进行傅里叶变换转换为频谱图
将频谱图的条纹二值化处理并进行最小二乘拟合
求取拟合直线斜率
实现运动方向的测量。根据所求点扩散函数(PSF)
利用Lucy-Richardson迭代算法对运动模糊图像进行复原。相对于传统方法
所提出的方法对模糊长度和模糊角度的确定具有较高的精度
对视觉传感器运动引起的模糊图像复原具有良好的效果。
Abstract
Aiming at problem that the jitter of visual sensor causes the captured image to be blurred
existing blur length and blur direction parameter identification methods is studied. According to correlation of moving blur images background pixels point
the Sobel operator is used to perform the first-order differentiation of blur images
and autocorrelation operation is carried out
two boundary values of each column of the correlation curve is summed
the distance between two equal minimum values in the summing curve is equal to the motion blur length.Blurred image is transformed to spectrum through the Fourier transform
binarize the stripes of the spectrum and then use least squares fitting to obtain the slope of fitted straight line
realize measurement of motion direction.According to the obtained point spread function( PSF)
recover motion blurred image by Lucy-Richardson iterative algorithm. Compared to traditional methods
the experimental results show that the proposed method has high precision for determination of blur length and blur angle and has good effects on the restoration of visual sensor motion caused blur image.