This paper propose a very simple feature detector which computes the mean, max and min of pixels over columns and rows in a square mask. Then the squared difference of the mean and max, and the mean and min, respectively are computed (for each row and each column) and stored in a feature vector. This very simply approach gives a rather surprisingly robust feature vector, despite its simplicity.

circle

A rotation invariant feature vector based on the same principles can be obtained by sampling in circles instead computing the same squared difference per circle, except for the centre, where that pixel value is stored.

  • A Simple and Efficient Feature Descriptor for Fast Matching. A. Hast, V. Sablina, G. Kylberg, I-M, Sintorn. WSCG2015, Full Paper. pp. 135-142. 2015
    @inproceedings{Has15a, author = {A. Hast and V. Sablina and G. Kylberg and G. Kylberg I-M and G. Kylberg I-M Sintorn}, title = {A Simple and Efficient Feature Descriptor for Fast Matching}, booktitle = {WSCG}, pages = {135--142}, year = {2015} }

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s