Eny

(a) Ground truth (b) DML (c) PVB (d) PPVC (e) Slice geometry

FIGURE 10 Comparison of DML classification (b), the PVB classification (c), and PPVC classification (d). (a) is a reference for what "ideal" classification should produce. Note the band of dark background material in (b) and (d) between the two curved regions. This band is incorrectly classified and could lead to errors in models or images produced from the classified data. The original dataset is simulated, two-valued data of two concentric shells, as shown in (e), with SNR of 14.2.

(a) Ground truth (b) DML (c) PVB (d) PPVC (e) Slice geometry

FIGURE 10 Comparison of DML classification (b), the PVB classification (c), and PPVC classification (d). (a) is a reference for what "ideal" classification should produce. Note the band of dark background material in (b) and (d) between the two curved regions. This band is incorrectly classified and could lead to errors in models or images produced from the classified data. The original dataset is simulated, two-valued data of two concentric shells, as shown in (e), with SNR of 14.2.

TABLE 1 Comparative RMS error for the four algorithms: PVB, PPVC, Mixel, and Eigen"

RMS error

- Improvement ratio

Background

0 0

Post a comment