The Research On Restoration Algorithm Of Motion-blurred Image

The restore algorithm of the image which is motion blurred and noised is proposed. First, the mathematical model based on motion blur system is constructed. The PSF of the motion blur in the form of frequency-domain is given,which is characterized by two parameters, blur direction and blur length. According to the characteristics of blurred images, the parameters of PSF are estimated. In fact, the modification of the cepstrum is given and the location of zero point of the specific spectrum (or close to zero value) in the spatial frequency plane is used to determine the parameters. Once the PSF is determined, many methods can be used for image restoration. The degradation function is constant throughout the image, while the edge of the image with rapid-changing gray value (with a smooth amplitude spectrum) plays more important role for estimating the parameters of the degradation function. Thus for a given motion-blurred image, we can select a few key areas for parameter estimation. Experiments show that the obtained point spread function by this method is more accurate. It can effectively reduce the errors caused by noise.