Abstract : A dynamic data driven application system prototype applicable to an adaptive multimodal sensor together with a proposed adaptive sampling strategy was demonstrated to greatly reduce the amount of data required to perform target identification. By using an adaptive sampling strategy, while approximately 10% of the image pixels were selected to collect spectral data to perform feature matching in the prototype examined here. Target identification was shown to be improved using a background data elimination method was designed to remove redundant spectral data and an adaptive forecasting strategy based on the extracted context from OpenStreetMap improved tracking accuracy and target identification. In summary, the vehicle tracking adjusts to the vehicle movement, the background environment, and the road network as derived from the imagery used in the tracking.
[1]
Bin Chen,et al.
Feature Matching and Adaptive Prediction Models in an Object Tracking DDDAS
,
2013,
ICCS.
[2]
Bin Chen,et al.
Novel spectral similarity measure for high resolution urban scenes
,
2012,
2012 IEEE International Geoscience and Remote Sensing Symposium.
[3]
Bin Chen,et al.
A Novel Adaptive Scheme for Evaluating Spectral Similarity in High-Resolution Urban Scenes
,
2013,
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.
[4]
John P. Kerekes,et al.
Adaptive Optical Sensing in an Object Tracking DDDAS
,
2012,
ICCS.
[5]
Bin Chen,et al.
Feature Matching With an Adaptive Optical Sensor in a Ground Target Tracking System
,
2015,
IEEE Sensors Journal.