TY - JOUR TI - Stereo vision-based vehicular proximity estimation DO - https://doi.org/doi:10.7282/T3BZ64HH PY - 2014 AB - This thesis describes an innovative and cost effective method to develop a low-cost following distance logging algorithm for volunteer participants which will allow quantitative research in driving behavior. Sparse stereo depth estimation methods along with a license plate localization algorithm has been used in order to achieve this. The depth is estimated by processing the video feeds from the stereo camera setup mounted inside a car looking out of the front window. License plate localization is used as a means to localize the position of the car within the image. Depth is estimated from the disparity, which is calculated using the rectified images of the frames from both the video streams. KW - Electrical and Computer Engineering KW - Motor vehicle driving KW - Image processing KW - Depth perception LA - eng ER -