Fig .3. Shows a sample of the evaluated values of S(x) with intensity as features F. The defined visual saliency measure S(x) gives out the contrast of the feature values between the B and K. Although visual saliency measure is related to other feature measures like the features that is discussed by Boiman and colleagues [8], but the main problem is that a posterior probability model is defined and will do the evaluation in semi-local windows, instead of one pixel at a time. By considering larger and bigger windows the algorithm will be able to be more independent to any model for the probability
