This is a project I am using to learn some more about C++ and computer vision.
The code is built on openCV 3.1.0
Use as laneTracker.exe path_to_input_video
The parameters for where the regions are located and what settings are used in the processing are hard coded right now. It is done in a way that should make reading from a file easy.
-A video stream is opened
-8 'regions of interest' are located based on some parameters (based roughly on the infinity point of the image)
-Each region is skeletonized
-A probablistic hough transform is done on the skeleton
-The results are drawn.