Graphics Reference
In-Depth Information
Fig. 16. Result of Karman Filter of the Distance
4
Experiments
In order to verify the performance of the practical application of the proposed method,
camera (controlled by a C++program?) is used to captured the video of the actual
driving. The vehicles in these videos are identified and located by using goal differen-
tial screening and coordinates mapping because the 1280 × 720 images on the PC can
handle about 10 frames per second. The results are shown in figure 17: there are total-
ly 396 frames in this segment of video, and according to the statistics, there are in
total 25 frames with detection errors; the error rate is 6.3%. The errors are mainly
due to the missing of detection, but the current distance between vehicles can still be
estimated by kalman filter when the vehicles are undetected (the results of vehicle
locating shown in figure 17). Thus, it will not have much impact on the locating of the
Fig. 17. Result of Vehicle Tracking and Location
Search WWH ::




Custom Search