CBIR Content Based Image Retrieval
Document Sample


CBIR
Content Based Image
Retrieval
Typical CBIR System
Similar Images Query Processor
I 4, I 8, … I k outputs
the list of images in
decreasing order of
similarity
F(Iq)
Attribute Comparison
using distance
Color histogram function
f(x)
Texture Desc
Shape details
Feature Extraction
Iq - Query Image
I1, I2, .. Image
Files
Applications
The need to retrieve images from a collection is there in
the following disciplines.
Satellite Image databases, Geographic Information System(GIS)
Architectural Design
Reconnaissance flights and Military
Crime prevention (Finger prints and face recognition)
Journalism and TV production
Art galleries and Museum Management
Advertising and Publishing
Fashion and Fabric design
Video on Demand
Medicine
E-Commerce and Intellectual Property (trade mark registration and
copy right management). ex: a trademark can only be assigned if it
does not resemble those in the database too closely.
Unfortunately not many from these disciplines know about CBIR.
User Query Example
Sorted results
Image
results of
the user’s
query
Results(continued)
Retrieval result examples:
• Example 1: query mode, using color features
(a) (b) (c) (d) (e)
Query mode result sample: (a) The query image. (b)-(e) The
similar image returned by the retrieval system. They are ranked
by the color similarity measurement, with (b) having the highest
ranking.
Airborne Laser Terrain Mapper
Components of LIDAR
• Laser Scanner (33-100 Pulses/sec)
• GPS receiver
• Inertial Navigation System
• The scanner records the difference in time
between emission of the laser pulse and the
reflected signal.
• The x,y,z coordinates of each ground point are
computed.
• Accuracy : 15 cms at 1000m flying height
Airborne Laser Terrain Mapper – Florence City
Airborne Laser Topographic Mapper – Power Line Sag Measurments
Get documents about "