Optionally, Automate NN Training with Classic Lane Tracking
Zur Navigation springen
Zur Suche springen
Classic Lane Tracking
Bird Eye view
The bird eye view transforms the normal camera angle to a top view of the track. This is the appropriate method to perform lane tracking, as seen from this project[1]. However, this project won't use birdeye view as the calibration parameters do not fully correct the distortion of the track, as seen below.
results
Edge detection lane tracking
Edge detection lane tracking uses Hough transform and Canny edge to detect lines, while the ROI (region of interest) reduces the surface of edge detection, which makes it more precise.
| # | Name | Description | Pictures |
|---|---|---|---|
| 1 | Double Edge Detection: | Double edge detection makes it difficult to deal with multiple angles of the track edges as seen in the image which detects the wrong lane. Therefore, single-lane detection ROI is more precise. |
  |
| 2 | Inverted image: | Inverting the image color makes the edge detection more precise. |
 
|