Melvin's digital garden

Albuquerque2004

CREATED: 200906170352 LINK: url:~/Modules/Literature/Albuquerque2004.pdf Summary: thresholding using Tsallis entropy

Shannon’s entropy: \[S = - \sum_{i=1}^k p_i \lg p_i\] \[S(A+B) = S(A) + S(B)\]

Tsallis entropy: \[S_q = \frac{1 - \sum_{i=1}^k (p_i)^q}{q-1}\] \[S_q(A+B) = S_q(A) + S_q(B) + (1-q)S_q(A) S_q(B)\]

Parameter q is an entropic index that captures the amount of nonextensivity (long time memories, fractal type structures). This expression meets the BGS entropy in the limit \(q \rightarrow 1\)

Thresholding equivalent to finding t which separate intensity values into two groups (object and background) that maximizes the Tsallis entropy

Links to this note