Blur identification and correction for a given imaging system

Estimating the blur function is the first step in all the image restoration techniques. A priori knowledge of the blur phenomenon helps improve the quality and certainty of image restoration. In this paper, a method of estimating the blur function for imaging setups that are not subject to change, for example surveillance cameras, is developed using the Wiener filter. A random noise image is used as a test image. Deconvolution is performed between the original test image and its observation taken from the given imaging system, to obtain the blur estimate for the system. Necessary preprocessing steps are implemented to compensate for the non-ideal nature of the imaging environment. Blur correction is then implemented on different observations taken from the same imaging system. The performance of the restoration is evaluated using mean square error and signal to noise improvement criteria.