Welcome to ROSA P | Commercial Vehicle Route Tracking Using Video Detection - 20477 | US Transportation Collection
Stacks Logo
Advanced Search
Select up to three search categories and corresponding keywords using the fields to the right. Refer to the Help section for more detailed instructions.
 
 
Help
Clear All Simple Search
Advanced Search
Commercial Vehicle Route Tracking Using Video Detection
  • Published Date:
    2010-10-31
  • Language:
    English
Filetype[PDF-1.98 MB]


Details:
  • Corporate Contributors:
  • Publication/ Report Number:
    OTCREOS9.1-16-F
  • Resource Type:
  • TRIS Online Accession Number:
    01349471
  • Edition:
    Final Report
  • Format:
  • Abstract:
    Interstate commercial vehicle traffic is a major factor in the life of any road surface. The ability to track these vehicles and their routes through the state can provide valuable information to planning activities. The authors propose a method using video cameras to capture critical information about commercial vehicles when they enter the state and store this information for later retrieval to provide tracking functions. As these vehicles continue on their routes, additional cameras will capture images that can be used for route tracking. By using these data, reports and highway utilization maps could be generated showing commercial vehicle routes and vehicle counts for state highways. Spurred by the competitive performance potential realized in face recognition via sparse representation, the authors treat the problem of vehicle identification with different video sources as signal reconstruction out of multiple linear regression models and use compressive sensing to solve this problem. By employing a Bayesian formalism to compute the l-1 minimization of the sparse weights, the proposed framework provides new ways to deal with three crucial issues in vehicle identification: feature extraction, online vehicle identification dataset build up, and robustness to occlusions and misalignment. For feature extraction, the authors use the simple down-sampled features which offer good identification performance as long as the features space is sparse enough. The theory also provides a validation scheme to decide if a newly identified vehicle has been included in the dataset. Moreover, unlike PCA or other similar algorithms, using down-sampling based features, one can easily introduce features of newly identified vehicles into the vehicle identification database without manipulating the existing data in the database. Finally, Bayesian formalism provides a measure of confidence for each sparse coefficient. The authors have conducted experiments to include different types of vehicles on the interstate highway to verify the efficiency and accuracy of their proposed system. The results show that the proposed framework cannot only handle the route tracking of commercial vehicles, but works well for all classes of vehicles.

  • Main Document Checksum:
  • Supporting Files:
    No Additional Files
No Related Documents.
You May Also Like:
Submit Feedback >