Research On Restoration Technology Of Motion Blur Image And Its Algorithm Realization

In a conventional single exposure time, relative motion between camera and photographed object can cause motion blur. Motion blur reduces image quality, and has a serious impact on people exploring the outside world, acquiring information and development of science and technology. Therefore, research on motion deblurring has very important significance, which widely used in industrial control, road monitoring, military, medical and forensic fields. The main work of motion deblurring is to remove or alleviate the degradation of image in image acquisition process, then makes the motion blurred image clear. This paper's research work is carried out from this point.Because the traditional shutter of the camera is equivalent to a low-pass filter, the motion blurred image loses its important high-frequency spatial details so that motion deblurring via deconvolution becomes an ill-posed problem. In this paper, first, according to the spectrum characteristics of motion blurred image and Hough transform, a algorithm based on the bright line of motion blurred edge image spectrum is presented to identify the direction of motion blur; second, according to analysis of the correlation of adjacent pixels along the direction of motion in the motion blurred image and optical flow equation, the length of motion blur is identified by difference autocorrelation function of motion blurred image; then the imaging principle of coded exposure is described, which using a binary code to control the camera's shutter open and down during the entire exposure time. This can be compared with the spread spectrum in communication system and the shutter is equivalent to a band-pass filter. The motion blurred image preserves high-frequency spatial details, and coded exposure imaging is simulated by mutiframe adding; finally, PSF is estimated according to the characteristics of coded exposure, and according to the parameters of motion blur, the clear image is restored based on MMSE. The quality of restored image is evaluated by the method of combining subjective and objective.Experimental results show that the methods presented in this paper can accurately identify the motion blur parameters and effectively restore the motion blurred image which captured based on coded exposure.