Please enable JavaScript.
Coggle requires JavaScript to display documents.
Multi-Sensor Data Fusion (JDL top level fusion process model 1987…
Multi-Sensor Data Fusion
JDL top level
fusion process model
1987
Source
Pre processing
Level One:
Object Refinement
Level Two
: Situation Refinement
Level Three
: Threat Refinement
Level Four
: Process Refinement
Human Computer Interaction
Database Management
Suppor
t Database
Fusion
Database
Types of
architectures
for data fusion
Locational
Fusion
Autonomous fusion
: Each sensor performs a single source positional estimation, producing a state vector from each sensor.. This reduces communication between sensors and fusion processor. not as accurate as data level fusion due to information loss.
Hybrid approach
: Fusion of raw or state data. Involves fusion of data.
Centralised fusion
: theoretically the most accurate way to fuse data, assuming association and correlation can be performed correctly. Typically usses estimation techniques such as the
kalman filter
.
Identity
Fusion
RAW or DATA LEVEL
: fusion of raw observational data. Can be directly combined if data is commensurate, otherwise will need to be done at feature/state level or decision level
Techniques
: Classic detection and estimation methods (sequential estimation techniques such as kalman filters)
FEATURE OR STATE LEVEL
: features are extracted from multiple sensor observations and combined into a single concatenated feature vector which is inputted to a
pattern recognition
approach
Pattern recognition algorithms
: Neural Networks, clustering algorithms, template methods.
DECISION LEVEL
: