Finding the lanes using histograms

How could we understand, more or less, where the lanes are? Visually, for a human, the answer is simple: the lane is a long line. But what about a computer?

If we talk about vertical lines, one way could be to count the pixels that are white, on a certain column. But if we check the image with a turn, that might not work. However, if we reduce our attention to the bottom part of the image, the lines are a bit more vertical:

Figure 3.25 – Lane with a turn, after the threshold, the bottom part

We can now count the pixels by column:

partial_img = img[img.shape[0] // 2:, :]  # Select the bottom part
hist = np.sum(partial_img, axis=0)  # axis 0: columns direction

To save the histogram as a graph, in a file, we can use Matplotlib:

import matplotlib.pyplot as plt

plt.plot(hist)plt.savefig(filename)plt.clf()

We obtain the following result:

Figure 3.26...