I would say to try to detect the road with a linear Hough transform, then extract the pixels that belong to the road. If it's done, you're halfway there.
Then, do a segmentation. You can use k-means or an adaptative thresholding to segment the "things" on the road. Do some erosion/dilatation to get rid of the noise. Now, you should have the silhouettes of the vehicles.
Finally, use a Kalman filter to track the movement of the objects.
Hope this helps.
Asked: 2016-01-26 00:54:18 -0600
Seen: 340 times
Last updated: Jan 26 '16
Please, put in some effort. We are not a
i need a solution, give me now
kind of forum. Take a look at the FAQ to fully understand my reaction ... This seems again like a school assignment and a student picking the shortest way to a solution...