Lane Detection based on Hough Transform

The video shows the results of a simple lane detection algorithm based on Hough transform. The lane is detected by finding the main lane markings on the roads. Specialized filters to find the single and double band lane markings are applied to a inverse perspective image of the road (bottom left). The inverse perspective image is integrated over time according to the motion of the vehicle. Band pass filters are applied to enhanced image positions were the lanes are to be found. Then, the lane markings are found by a specialized Hough transform that finds parallel line markings (second column) on the resulting enhanced filtered images (third column) Finally, a simple Kalman filter estimates the position of the lanes as the vehicles drives. The computational complexity of the algorithm is very low allowing over-real-time lane detection and tracking. Please, checkout my computer vision development framework QCV at SourceForge.net http://sourceforge.net/projects/qcv/ .