Performance Evaluation Of Packet Thresholding For Image Denoising
Image denoising plays a vital role in the field of image processing. Wavelet transform based denoising is a popular
approach since last decade. In this paper, wavelet packet transform based Image denoising is evaluated. Evaluation is done on
the basis of performance of wavelet packet thresholding algorithm for various noise levels of various noises and for db8 and
haar wavelet. We propose that use of Median filter and Wiener filter after wavelet packet thresholding improves PSNR
depending on the type of noise respectively.
Experimental results for 256 X 256 gray images indicate that the packet thresholding algorithm is best suited for Poisson noise
compared to Speckle, Gaussian, Salt and Pepper noise. Filtering after WPT improves performance of the algorithm but at the
cost of computational complexity. Performance of the algorithm decreases as noise level increase.