Image Processing

If p(x, y) were known completely, subtracting p(x, y) from g(x, y) will give f(x, y). In practice, only an approximation of the true pattern can be obtained. The effects of the components that are not considered in the estimation of p(x, y) can be minimized by subtracting g(x, y) by weighted p(x, y), which is given by

$\widehat{f}\left(x,y\right)=g\left(x,y\right)-w\left(x,y\right)p\left(x,y\right)\left(4.103\right)$

where w(x, y) is a weighting function. The weighting function w(x, y) can be selected to get optimal results under certain conditions. For example, one approach is to select w(x, y) so that the variance of (x, y) is minimized over a specified neighborhood of ...