Intelligent Image Restoration Approach: Using Neural Networks to Eradicate Dilemma in Punctual Kriging

We report an intelligent image restoration approach by combining the geostatistical interpolation technique of punctual kriging and the machine learning approach of adaptive learning. Digital images degraded from Gaussian white noise are restored by first utilizing fuzzy logic for selecting pixels that need to be kriged. The concept of punctual kriging is then used to estimate the intensity of a pixel. Kriging un-biased estimates mostly suffer from occurrence of negative weights and matrix inversion failure problems. Approximation is usually used to avoid these problems in punctual kriging based image restoration. Artificial neural networks (ANN) are employed to minimize the cost function of the kriging based pixel intensity estimation procedure. ANN, in merit to analytical methodologies, avoids both matrix inversion failure and negative weights problems. Experimental results using four hundred and fifty images and different image qualitative measures show the superiority of the proposed method against adaptive Weiner filter and existing fuzzy kriging approaches. This also validates the use of hybrid approaches to image restoration problem. [Chaudhry A, Khan A, Kim JY, Niu QQ. Intelligent Image Restoration Approach: Using Neural Networks to Eradicate Dilemma in Punctual Kriging. Life Sci J 2013;10(1):1631-1641] (ISSN: 1097-8135). http://www.lifesciencesite.com. 240