Advanced Lane Finding

An updated image/video processing pipeline using advanced computer vision fuctions to identify lane bondaries from dashcam recordings.

The following key concepts are explored:

• Camera calibration
• Distortion correction
• Color spaces and gradients
• Perspective transform and recovery
• Polynomial fitting
• Pixel-to-meter conversion and calculation
• Sliding window and histogram
• Weighted moving averaging

An example of a raw camera image is shown in below, as well as a processed image with lane boundaries detected. The radius of the curvature and the vehicle position relative to the center of the road are displayed onto the image as well.

You can view an example video output with lane bondaries detected in below.

For details, please follow this link.