CORRECTION OF A NOISY IMAGE BY A POLYNOMIAL APPROACH AND CHOICE OF THE BEST IMAGE BY ONE OF THE POLYNOMIAL’S ROOTS.

Abstract

In this paper a polynomial method of selecting an image disturbed and corrected by the modified power law by one of its roots, is proposed. This power law uses here is a real power variable belonging to the interval [1.00,..1.12]. It provides a dozen corrected images. But it is difficult to get the best image between them, or the image which has the best signal to noise ratio. One of the roots provides this value. Comparison of reconstructed image with the original is proved by structural similarity index (SSIM), entropy and peak signal-to-noise ratio (PSNR) which are objective quality measures and the averages of gray levels of pixels which are very similar. The polynomial selection method has the advantage of providing only a single corrected image without RGB YCbCr transformation noise and close to original among many others. Where somebody needs to choose one image among several, this method can provide solution.