Figure 5

Color image with vector geometries. Graphical representation of two orthogonal eigenvectors on a current point p (a). Some two orthogonal eigenvectors depicted (b), vector edge indicator function g = (1 + λ⁺λ^-)^-1(c), vector gradient norm calculated by$\sqrt{{\lambda }^{+}}$(d), vector gradient norm calculated by$\sqrt{{\lambda }^{+}-{\lambda }^{-}}$(e), and vector gradient norm calculated by$\left|\right|\nabla \mathbf{I}\left|\right|=\sqrt{\mathrm{trace}\left(\mathbf{K}\right)}=\sqrt{{\displaystyle \sum _{i=1}^3}\left|\right|\nabla I_i\left|\right|}=\sqrt{{\lambda }^{+}+{\lambda }^{-}}$(f).