Pixel-Based Statistics and a Probabilistic Representation
Jason W. Trobaugh, R. Martin Arthur

The physical model provides the mathematical basis for computing a pixel-based probabilistic representation for the image. We compute the mean, variance and SNR0 at each pixel, resulting in statistical images as shown. 

From the statistical images, we classify each pixel as Rayleigh or Gaussian, depending on the value of SNR0. The data likelihood, a probabilistic image model, is then constructed by assuming independence of neighboring pixels. Samples of the resulting image model are also shown.

For more information, contact us or see the Publications. 

Statistical Images
m
s
SNR0
Samples of Rayleigh/Gaussian Image Model

 
Home
Information/Contact  Publications  Collaborators Related Links