adaptive (local) thresholding

Giving every pixel its own threshold, computed from its own local neighborhood, instead of one shared global value.

Adaptive (local) thresholding computes a per-pixel threshold $T(x,y) = \bar f_{N(x,y)} - C$ , the mean of a local neighborhood $N(x,y)$ around $(x,y)$ minus a constant $C$ , then classifies $g(x,y)=255$ if $f(x,y)\le T(x,y)$ else $0$ . Because a pixel's own value is always part of its own neighborhood mean, the foreground-below-threshold polarity is required for the scheme to behave sensibly: in a flat region the mean equals the pixel's own value, so only a pixel sitting meaningfully BELOW its surroundings (by more than $C$ ) is ever flagged.