Figure 1. Gaussian weighted local variance computation process.

A) During the process, the source pixel is replaced with a Gaussian weighted variance of all the pixels inside the chosen window size. Each pixel integer value in the source image is multiplied by the corresponding value in the overlying Gaussian kernel, and the variance of all the resulting products is computed. The gray value of the source pixel (P_x,y) is then replaced by this Gaussian weighted local variance. This operation is repeated for each pixel in the original image. B) Source image and C) corresponding output image after application of the algorithm.