This code uses Hough Transform to find highway road lane lines from the results of Canny Edge Dection.
source activate carnd-term1jupyter notebook P1.ipynb
1. Describe your pipeline. As part of the description, explain how you modified the draw_lines() function.
My image processing pipeline consists of the following steps:
- Convert image to grayscale
- Apply Canny edge detection
- Apply a Region of Interest Mask
- Use Hough Transform to Find Lines from Canny Edges
- Find the best fit line of the endpoints of the results of the Hough Transform
- Draw lines on a blank image of the same size.
- Draw the lines onto the original image
In order to draw a single line on the left and right lanes, I modified the draw_lines() function by:
- Grouping the lines into two categories (right and left) by checking if the slope was positive or negative. As well as throwing out any lines without an "extreme" slope.
- Using numpy's polyfit and poly1d functions I find the best fit line with the points for both the right and left sides individually.
One potential shortcoming would be what would happen when there is a different image format than the given input. My algorithm is very dependent on knowing the placement of the camera and the layout of the image.
Another shortcoming could be the dependence on clear lane lines in the image. In real road conditions the lane is not always perfectly marked. Old lane lines, road debree, other car and objects could obstruct the view of the lines and cause the program to give incorrect output.
A possible improvement would be to use a more advanced technique for finding the lanes so that the placement is not so dependent on camera placement.
Another potential improvement could be to create a more sophisticated algorithm for extrapolating the lane lines given the photo's context. You could use previous photos seen or the placement of other cars as well as the width of your car compared to with width of the road to better determine where the lanes might be if they are not visible.