Doubt about Speeded-Up Robust Features paper

edit

In the paper entitled "Speeded-Up Robust Features (SURF)" Bay et. al. mention that

... ‘interest points’ are selected at distinctive locations in the image, such as corners, blobs, and T-junctions.

Can someone explain to me what they mean by T-junctions?

I know this question is not directly related to the OpenCV library but as the OpenCV implements the SURF algorithm, maybe someone who uses it can explain it to me.

Thanks in advance

Reference:

Bay, Herbert, et al. "Speeded-up robust features (SURF)." Computer vision and image understanding 110.3 (2008): 346-359.