Novel Heuristics for Deconvolution Applied to Picture Deblurring
Evgeniya Olenuk, Maxim Gromov · Proceedings of the Spring/Summer young researchers' colloquium on software engineering · 2012
In this paper we describe an implementation of a method for blurred pictures restoring.We consider uniform, linear photo-camera shift (movement) while picture is taken.In this case, a picture, formed on a chip of a camera, is not static, but moving along some straight line, and as a result photo is blurred.Blurred picture can be seen as a convolution of two functionsone describes initial picture, another is point-spread (or blurring) function (PSF) [1]-[3]."Deblurred" picture can be found as a deconvolution of the given picture with known PSF [1], [2].To do this it is convenient to use Fourier transform and apply the convolution theorem.But it leads to division by 0, and thus the method is unstable.We suggest two heuristics to avoid division by 0 and compare them with the known one [2].