Directional Differences Clause Samples

Directional Differences. In order to account for the large morphological variation of the images in our dataset, we evaluated differences between pixels in various directions. Consider two pixels p1 and p2 on a two-dimensional image I(x, y). Assume that the Euclidean distance between p1 and p2 be d and p2 lies on a line that has an orientation of angle α with respect to the horizontal coordinate: I(x + d cos α, y + d sin α)Let us construct a vector of size i = 1, 2, ..., A as follows: Iα = [Iα(1), Iα(2), ..., Iα(A)], (8) where Iα(i) = I(x+i· R cos(α), y+i· R sin(α)), α is a spec- (a) BT-20 class (b) Cama-1 class (c) HepG2 class class 1. Examples of the image classes used in our experi- ments.