Deconvolution of noisy images

By seeking the least detailed picture that is consistent with a given set of data, the maximum entropy method is capable of increasing the contrast of an image, and can also increase its sharpness if the measurements are sufficiently accurate.

"Susie" -- original image

High-resolution low-noise original image
Susie -- blurred, high noise. Restored image of Susie when noise is high
Blurred original with high added noise.
 Maximum entropy suppresses noise.
Susie -- blurred image, low noise Restored image when noise is low.
Blurred original with low added noise Maximum Entropy de-blurs only if data are accurate enough.

Back Back to list of examplesBack to previous exampleNext example Search | Home

©MEDC Ltd.
Last revised: September 19, 1999

Comments/corrections