Research Of The Motion Blurred Image Restoration In Printing Detection

In the full screen print test, due to the stillness of the CCD and rapid detection of the printing products lead to relative movement blurred, influence subsequent defect detection, when the fuzzy degree serious and even can’t detection, error warning, etc., aiming at the problem in this paper, we study the motion blur image restoration method, through the establishment of degraded image restoration model to achieve the restoration of blurred image, and finally applied to the full image detection system.Take different filter parameters comparative median filter and mean filter respectively to the removal efficiency of different noise density image, the median filter is obtained for the blurred image denoising effect is more effective. And according to the inverse process of original image backward image degradation, namely, according to the fuzzy image and point spread function (PSF) restored image. PSF system of point spread function is unknown, there are two important parameters of point spread function (PSF) that are motion blur length and motion blur direction, for the only two reference input to the filter can restorat image.Analysis two kinds of calculating method of blured direction, one is Fourier spectrum method, another methods is direction of differential. By comparing different blured length corresponding to the different Angle respectively to calculate the accuracy of the two methods, analysis the range of allowable accuracy can use direction differentiation, high precision should use the Fourier spectrum method.The autocorrelation function based on sobble operator, and can conclude that the autocorrelation function of image on the absolute value of the horizontal distance between the two negative peak is the motion blur length, by this method can accurately calculate the length of motion blur. Be particular about the length and Angle parameters of the input to the filter for image restoration.Four restoration principles of blind deconvolution, least square method,wiener filtering, and Richardson-Lucy are explained, doing restoration experiment saiming at every method on condition of noise and no noise, experiment results show the restoration effect of Lucy-Richardson and use in print testing are both better than other methods.Part of the image restoration experiment using Matlab programming, respectively to calculate the PSF of blur parameters can contrast, image denoising, four kinds of recovery filter of image restoration and the effect of restored image evaluation.The application of the restored image through the full picture of defect detection system based on vc++.