Wavelet based Denoising for Images Degraded by Poisson Noise

T. Shimamura, S. Eda, T. Ito, Y. Kuwano, and Y. Takahashi (Japan)


Poisson noise, medical image, denoising, wavelet shrink age, pixel values based division, SPECT


This paper presents a method to denoise an image degraded by Poisson noise. Poisson noise is signal-dependent so that the variance of the noise changes in proportion to pixel val ues of the image. In the method, the pixel values based division technique, which was recently addressed, is first utilized to whiten Poisson noise. Then, the wavelet shrink age is used to remove the remaining noise in the processed image. Simulation experiments demonstrate that the pro posed method works well on SPECT images.

Important Links:

Go Back