Simple computer vision project for hackaton, created in 1 day. Based on OpenCV.
For given input like this:
The image is converted into grayscale, histogram is equalized and subtle gaussian blur is applied:
Then Canny edge detection algorithm is used and image is cropped to get this:
Finially, Hough transform is performed, to detect the lines:
Algorithm evaluation is based mainly on the angles between edges of the road and x-axis, look at the source code for more details.