Remote-Sensing Image Tile-Level Annotation Based on Discriminative Features and Expressive Visual Word Descriptors

Table 1

The UA value comparison between using different methods.

Methods	Water	Vegetation	Residential area	Farmland
Color features	0.871 2	0.955 9	0.731 8	0.696 4
Texture features	0.666 1	0.931 3	0.343 1	0.611 0
Spatial features	0.936 6	0.886 6	0.552 1	0.746 4
NOF + K-means	0.952 4	0.949 3	0.723 6	0.788 2
WOF + K-means	0.954 0	0.951 7	0.763 4	0.796 6
Our proposed method	0.957 4	0.957 8	0.780 8	0.818 5