Identification of Blur Parameters from Motion Blurred Images

the smear extent of the blurred image of a point object The problem of restoration of images blurred by relative in the original image. Extraction of the blur extent has motion between the camera and the object scene is important significant meaning in identification of the motion-blur in a large number of applications. The solution proposed here PSF. Cannon [1] dealt with the case of uniform linear identifies important parameters with which to characterize the motion blur that is described by a square pulse PSF and point spread function (PSF) of the blur, given only the blurred used its property of periodic zeros in the spectral domain image itself. This identification method is based on the concept of the blurred image. These zeros were emphasized in that image characteristics along the direction of motion are the cepstral domain and the blur extent was estimated by different from the characteristics in other directions. Depending measuring the separations between the zeros. The assumpon the PSF shape, the homogeneity and the smoothness of the tion of zeros in the spectral domain is not satisfied in blurred image in the motion direction are greater than in other directions. Furthermore, in this direction correlation exists bevarious cases of motion degradation such as accelerated tween the pixels forming the blur of the original unblurred motion [2, 3] and low frequency vibrations [4]. objects. By filtering the blurred image we emphasize the PSF Recent important developments in image restoration characteristics at the expense of the image characteristics. The are the maximum likelihood image and blur identification method proposed here identifies the direction and the extent methods [5–7]. These methods model the original image, of the PSF of the blur and evaluates its shape which depends on the blur, and the noise process. The original image is modthe type of motion during the exposure. Correct identification of eled as a two-dimensional autoregressive (AR) process, the PSF parameters permits fast high resolution restoration of and the blur is modeled by a two-dimensional linear system the blurred image.  1997 Academic Press with finite impulse response. A maximum likelihood estimation is used for identification of the image and blur parameters. The identification of the blur model parame