KLT tracker: vision.PointTracker is whether affine model or translation model

I would like to know whether

The implementation of KLT tracker in computer vision toolbox is the old KLT paper implementation (Only assumes Translation motion) or the ShiTomasi paper (Assumes affine model: 'good features to track').

Answer 1

[Pratyush Roy]

Hi Nisha,

The KLT Algorithm uses the 'good features to track' which assumes the affine model.

You can go thourgh the following documentation link for futher help:

https://www.mathworks.com/help/vision/ug/face-detection-and-tracking-using-the-klt-algorithm.html

[Nisha Varghese]

Thank You Pratyush. :)