next up previous contents index
Next: fsize Up: Reference Previous: fpsnr255   Contents   Index

frthre

$ \bigcirc$Name


frthre Thresold the pixel's gray-levels of a noisy fimage for optimal recovery




$ \bigcirc$Command Synopsis


frthre [-l noise] fimage result



-l noise : Absolute value of the additive noise

fimage : Input fimage

result : Output image




$ \bigcirc$Function Summary


void frthre (A , B , noise )

Fimage A , B ;

float *noise ;




$ \bigcirc$Description


frthre copies the input image A in the output B and makes a thresold of the pixel's values according the rule:

B(x, y) = $\displaystyle \left\{\vphantom{
\begin{array}{ll}
0 & \mbox{ if } \vert A(x,y)...
...x{sgn}(A(x,y)) \times \mbox{\tt noise} & \mbox{elsewhere.}
\end{array}}\right.$$\displaystyle \begin{array}{ll}
0 & \mbox{ if } \vert A(x,y)\vert \leq \mbox{\...
...) - \mbox{sgn}(A(x,y)) \times \mbox{\tt noise} & \mbox{elsewhere.}
\end{array}$

Where noise is the absolute maximal value of the noise (noise level).

In [DJ94], it is proved that this simple thresold rule is one of the best estimator for recovering a noisy image (additive noise) using the MiniMax error, when the coefficients in A and B are coordinates in an orthonormal basis (such as wavelet basis).




$ \bigcirc$Version 1.0


Last Modification date : Thu Nov 29 20:23:56 2001


$ \bigcirc$Author


Jacques Froment






next up previous contents index
Next: fsize Up: Reference Previous: fpsnr255   Contents   Index
mw 2004-05-05