This repo contains the code written to complete the first project on Udacity Self-Driving Car Nanodegree Program (Term 1). This project aggregates algorithms to identify and mark lane lines on the road. The algorithms work for images and videos which are taken from a camera at the center of a vehicle.

The pipeline consists of six steps:

1. Converts the image into a gray scaled version using the cv2.cvtColor() function.
2. Applies a Gaussian blur to the image using the cv2.GaussianBlur() function.
3. Uses a Canny transformation to find edges on the image using the cv2.Canny() function.
4. Applies an image mask to keep only the region of interest using the cv2.fillPoly() function as well as the cv2.bitwise_and() function.
5. Uses a Hough transformation to find the lines on the masked image using the cv2.HoughLinesP() function. It also draws these lines onto a black image (all black) of the same size as the original image using the cv2.line() function.
6. Merges the output of the last step with the original image to represent the found lines on it using the cv2.addWeighted() function.

In order to draw a single line on the left and right lanes, I modified the draw_lines() function (part of step 5) by adding a decision whether a line segment is part of the left lane or the right lane. The decision contains two parts. First, the slope ((y2-y1)/(x2-x1)) must be either positive or negative. Second, the positions of x1 and x2 must both be either on the left or on the right side of the image. The functions numpy.polyfit() and numpy.poly1d() are being used to average and extrapolate the line segments.

Step	Output
1
2
3
4
5
6