Figure 1.
General flowchart of proposed approach (KC: Kappa Coefficient; TC: Time-Consuming).
Figure 2.
Flowchart of the iterated procedure used to determine the key parameters ( and
) (KC: Kappa Coefficient; TC: Time-Consuming).
Figure 3.
Selection process of portfolio optimisation model (KC: Kappa Coefficient; TC: Time-Consuming).
Figure 4.
Location of the study area: Weichang County, Hebei Province, China.
Table 1.
The results of KC (Kappa Coefficient).
Figure 5.
Relationship among ,
and TC (Time-Consuming, unit: second).
Table 2.
TC (Time-Consuming, unit: second) of the proposed methodology with different parameters.
Figure 6.
Comparison of land cover classification in Weichang.
Table 3.
Error matrix of the combination model (80%, 80%).
Table 4.
Error matrix of the combination model (20%, 20%).
Table 5.
Error matrix of the combination model (60%, 60%).
Table 6.
Error matrix of common classification approach (Maximum Likelihood Approach).
Table 7.
Comparison of the Z values in each model/approach.
Figure 7.
Sketch map of automatic dataset of pure-pixel training samples.