Highway Lane Detection

Highway Lane Detection was a course project in the ENPM673 Perception for Autonomous Robots course. The project aimed to detect straight and curved lanes on a highway to be used as a sub-system on a driverless vehicle. 

 

The algorithm used is as follows:

  • Read image

  • Convert to grayscale

  • Perform Sobel Edge Detection

  • Perform Morphological Operations to make edges more prominent and remove noise

  • Obtain processed image using a combination of the resulting image along with color detection (based on yellow and white) 

  • Extract the Region of Interest from the processed image

  • Extract Hough Lines from the resulting image based on certain conditions and perform curve fitting on the three most prominent lines to estimate the driving lanes

 

 

 

The project source code can be found on Github at: https://github.com/bharatm11/Lane-Detection

 

RGB.jpg

Original

Original

grayscale.jpg

Grayscale

cleaned_edges.jpg

Clean Edges

Clean Edges

color_mask_im.jpg

Color Mask

Color Mask

fin_processed.jpg

Final Processed Image

Hough_lines.jpg

Hough Lines