Cryptography Reference
In-Depth Information
Fig. 11.22.
The retrieval interface of our system.
Table 11.2.
The average recall and precision for each class.
Class number Semantic Recall Precision
1
People
0.27
0.02
2
Beach
0.14
0.01
3
Building
0.20
0.01
4
Bus
0.24
0.02
5
Dinosaur
0.67
0.29
6
Elephant
0.35
0.02
7
Flower
0.24
0.04
8
Horse
0.48
0.12
9
Mountain
0.18
0.02
10
Food
0.20
0.01
11.4.5 Conclusions
This chapter outlines an image retrieval system based on a multipurpose wa-
termarking scheme. The advantages of this scheme are in three areas. The
first is that the system embeds the features in the images, and no extra space
is required to save the feature data. The second is that the image file can
be copied to other database without recomputing the corresponding features.
The thirdly is that the image is copyrighted and annotated, and the retrieval
system can confirm that the image and the semantic are original.
