A Study on the Effect of Gaussian Noise on PSNR Value for Digital Images
Parminder Kaur, Jagroop Singh · International Journal of Computer and Electrical Engineering · 2011
Most of the rules relating simple nonlinear threshold values for wavelet-based de-noising have assumptions that the wavelet coefficients are independent values. However, when we talk of natural images, we observe that wavelet coefficients have significant dependency. The phrase Peak Signal to Noise Ratio, often abbreviated PSNR, is an (9) engineering term for the ratio between the maximum possible power of a signal and the power of corrupted noise that affects the fidelity of its representation. In this paper, experimentation is performed to study the effect of Increasing Gaussian noise on PSNR and the corresponding measure. The last section of the paper illustrates the above explained concept with the graphical view of PSNR with increase in noise. I. I NTRODUCTION There are various methods to remove noise from the image and hence one can increase Peak Signal To Noise (PSNR) ratio. Some prominent work proposed by (1) as implement de-noising method by using the dual-tree complex wavelets into the ordinary ridgelet transform. The paper (1) used shift invariant property of the dual-tree complex wavelet and the high directional sensitivity of the ridgelet transform. In that paper, digital complex ridgelet transform is applied for de-noising of some standard images embedded in white noise. Here, it used hard thresholding of the complex ridgelet. Experimental results show that the new method is better than VisuShrink, the ordinary ridgelet image denoising. The paper (1) used wiener filter that is available in the Matlab Image Processing Toolbox. Complex ridgelets could be applied to curvelet image de-noising as well. In paper, (2) a better method is described for removing additive white noise of known variance from photographic images. The technique is based on a characterization of statistical properties of natural images represented in complex wavelet decomposition. Specially, the paper decompose the noisy image into wavelet sub-bands or patterns, estimate the autocorrelation of both the noise-free raw coefficients and their magnitudes (within each sub-band or pattern), impose these statistics by projecting onto the space of images having the desired autocorrelations. It reconstructs an image from the modified wavelet coefficients. De-noising results compare favorably to three reference