Auto-Focusing Algorithms

a) Absolute Gradient

b) Threshold Absolute Gradient

By summing those differences larger than a certain threshold value, only those larger gradients representative of the focused situation are taken into account.

c) Squared Gradient

By squaring, the algorithm enhances the larger gradients over the smaller ones. The advantage of this function over method c) is that it has no threshold to adjust.

d) Laplacian

In the fourier domain, the transfer function of a 2nd order diference filter (laplacian) suppresses more strongly the lower spatial frequencies than the first order difference filter (gradient).

e) Threshold Video Signal Content

Originally devised for microscopes. According to researchers, the threshold setting affects only the shape of the extremum, not its position.

f) Threshold Video Signal Pixel Count

Count the number of grey values above the threshold setting.

g) Signal Power

h) Standard Deviation

i) Normalised Standard Deviation

Divide the equation for h) by the mean of g(i,j). Division by the mean compensates for changes in the average image brightness.

j) Absolute Variation

Simplier than the standard variation method

k) Normalised Absolute Variation

Divide the equation for j) by the mean of g(i,j).


Main Page     Experiment