01472173

Component

Data on vehicle movements can be used to derive parameters for microscopic traffic flow models and calibrate traffic simulation models. A software tool called Tracking and Registration of Airborne Video Image Sequences (TRAVIS) has been developed to extract vehicle positions from airborne imagery to assist in analyzing microscopic traffic behaviors. One of the critical challenges in analyzing airborne image sequences is the detection of vehicles and continued tracking of the vehicles through the sequence. The proposed technique includes several enhancements to TRAVIS to improve the vehicle detection and tracking. Firstly, a lower threshold is used in a difference image, allowing for detection of pixels with lower contrast to the background. This step is critical to picking up vehicles with little contrast with the road color (RGB intensity). Secondly, different techniques are used to threshold the red pixels, normal pixels and the pixels with lower contrast to the background, based on: (1) the pixel intensities in the original image; (2) the pixel intensities in the differenced image; and, (3) the relative positions of vehicles on the roadway. Thirdly, the dilation step is removed to avoid merging the vehicles very near each other. Fourthly, two blobs belonging to one vehicle are merged when labeling the connected components. Finally, non-vehicle and stationary blobs are screened out. Experimental results show that the number of vehicles detected and tracked is significantly increased, compared to the previous method.

This paper was sponsored by TRB committee AHB15 Intelligent Transportation Systems.

01470560

13-2738

English

16p

2013

Transportation Research Board 92nd Annual Meeting

Digital/other

Figures; References; Tables

Contrast; Detection and identification systems; Image processing; Tracking systems; Vehicle detectors

Data and Information Technology; Highways; I70: Traffic and Transport

Transportation Research Board Annual Meeting 2013 Paper #13-2738

TRIS, TRB, ATRI

Feb 5 2013 12:34PM