26-08-2014, 11:39 AM
Comparative Study of X-Ray Image Denoising Using Wavelet Analysis Seminar Report
Comparative Study.pptx (Size: 2.39 MB / Downloads: 10)
Abstract
Image suppression and edge enhancement are performed based on the wavelet transform. The better denoising result depends on the degree of the noise ,generally its energy distributed over the low frequency band while both its noise and details are distributed over high frequency band and also applied threshold in different scale of frequency subbands to limit the image. To overcome the problem and to achieve better results this paper is proposed to indicate the Comparative Study of X- ray Image Denoising using Wavelet Analysis. Finally it compares the wavelets to produce the best denoised digital x ray image in terms of PSNR,MSE & EPI. Resulting image has high Peak signal to Noise Ratio (PSNR) ,minimum Mean Square Error (MSE) and high Edge preservation Index (EPI) value ,hence prove the robustness and effectiveness of the method .
Applications
Detecting Discontinuities and Breakdown Points
The discontinuous signal consists of a slow sine wave abruptly followed by a medium sine wave.
The 1st and 2nd Level Details (D1 and D2) Show the Discontinuity Most ClearlyThings to be Detected:
The site of the change
) The type of change (a rupture of the
signal, or an abrupt change in its first
or second derivative)
The amplitude of the change
Problem Identification
Some major challenges in the field of Wavelet analysis are:-
Analysis or Sounding out the local signal regularity
on a time scale
Denoising or Estimation of Function
Compression of Images
Principle of Denoising
Denoising consists of restoring a useful signal from observations corrupted by an additive noise.
The basic algorithm of denoising consists of :-
Decomposition
Selection or thresholding of coefficients
Reconstruction
Denoising
In the first column we see the wavelet coefficients from level 5 to level 1. To make them more readable they are repeated 2k times at the level k (which explains the gaps particularly visible for k > 3 ). In each of these graphs we note the presence of two horizontal dotted lines: the coefficients inside the band are zeroed by denoising.
In the second column the noisy signal s is superimposed on the denoised signal. Below there are two graphs: a colored version of the wavelet coefficients from level 1 to 5 of the original disturbed signal, followed by the counterpart for the thresholded wavelet coefficients, from which the denoised signal is reconstructed
Experimental Result
From Table No 4(a) : -MRI SCAN IMAGES
It can be seen that the Symlets Wavelet help in denoising , gives high PSNR value (66.9337), Minimum Mean Square Error (0.0131) and Edge Preservation Index equal to (10.1491)
From Table No.4(b) :-ULTRASOUND IMAGES
It can be seen that the Daubechies Wavelet help in denoising , gives high PSNR value (66.1488) , Minimum Mean Square Error (0.0157) and Edge Preservation Index equal to (6.6814).
From Table No.4© :- CTSCAN IMAGES
It can be seen that the Coiflets Wavelet help in denoising , gives high PSNR value (65.6549), Minimum Mean Square Error (0.0176) and Edge Preservation Index equal to (10.2101).
From Table No.4(d) :-X-RAY IMAGES
It can be seen that the Biorthogonal Wavelet help in denoising , gives high PSNR value (69.8478), Minimum Mean Square Error (0.0067) and Edge Preservation Index equal to (8.4285).
Conclusion
The proposed method is of low complexity ,both in its implementation and execution time. It adapts itself to unknown noise distribution and to the local spatial image contrast. We demonstrated its usefulness for noise suppression in Ultrasound, MRI,CT Scan& X-Ray images. The algorithm is not only enhancing the image contrast, but can preserving the original image quality