The transmitted gray-scale value gi, j of pixels i = 1,c, m, j = 1,c, n, in a digital space satellite photo20 is distorted by both the usual random noise and a known problem with the video camera that effectively multiplies the value for pixel (i, j) by a blurring factor bi, j. Engineers want to restore the image by estimating correct values for each pixel in a way that minimizes the total absolute deviation between predicted

(after blurring) and observed gray-scale numbers.

Formulate an LP model of this image restoration problem using decision variables 1i = 1,c, m; j = 1,c, n2 xi, j ! correct value for pixel 1i, j2 di, j

+ ! positive deviation of predicted over observed value for pixel 1i, j2 di, j

- ! negative deviation of predicted below observed value for pixel 1i, j2