For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Parameter Estimation and Restoration for Motion Blurred Images
IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences
Publication Date: 1997/08/25
Print ISSN: 0916-8508
Type of Manuscript: Special Section PAPER (Special Section on Digital Signal Processing)
restoration, motion blur, power spectrum, parameter estimation, Wiener filter,
Full Text: PDF>>
The parameter estimation problem of point spread function is one of the most challenging and important task for image restoration. A new method for the parameter estimation in the case of motion blur is presented here. It is based on the principle that the power spectrum of the motion blurred image contains periodical minima relevant directly to the motion derection and length. Though the principle is very simple and effective in certain cases, the direct use of it may lead to poor performance an the signal-to-noise ratio (SNR) gets lower. To improve the estimation accuracy, by analyzing image noise effect on the detection of the minima, we propose a method to greatly reduce spectral noise, and give the lowest allowed SNR at which the minima may still be identified reliably. We also estimate the power spectrum of the original image, which is a must for the Wiener restoration filter, from the noisy blurred image based on a noncasual autoregressive model. Once above parameters are decided, the Wiener filter is used to restore the noisy blurred image. Our method is very practical; no parameter needs to be known a priori, or to be adjusted manually to fit into various application problems. The proposed method is finally applied to systhesized and real motion blurred images to demonstrate its effectiveness.