How can I find the Sensitivity Factor of the Harris Corner Detector?

Hi,

I'm using the detectHarrisFeatures function. To learn about it, I've found this page:

http://it.mathworks.com/help/vision/ref/cornerdetection.html

If I well understand, it is about the Simulink Block for Corner Detection, is this right?

In this page, concerning the Harris Detector, I can read this: "The variable k corresponds to the Sensitivity Factor". So, which is default value of k used in the detectHarrisFeatures function? How can I find and edit it?

Thanks for your answers.

Answer 1

[Swarooph]

Yes you are right that the documentation link you have linked talks about the Simulink block and not the function detectHarrisFeatures function. There seems to be a difference in the way these 2 are implemented. The Simulink block uses the function cornerMetric under it. When I refer to the documentation for this function : http://www.mathworks.com/help/images/ref/cornermetric.html, I see that it is not recommended to use this and to use detectHarrisFeatures or detectMinEigenFeatures and the cornerPoints class in Computer Vision System Toolbox™ instead. So you seem to be on the right track. Just use the the detectHarrisFeatures function's documentation, you can find it here: http://www.mathworks.com/help/vision/ref/detectharrisfeatures.html. When you refer to the properties of the algorithm here. You are able to change the parameters MinQuality, FilterSize and ROI. Do read the documentation to see how to change the parameters to affect your detection.

[Nut]

Hi Swarooph, thanks for your answer, especially about the Simulink block. About the function, currently I use it very often, and I already know the documentation and what change to affect the detection. My question was not for a practical purpose, rather for a deeper understanding of the algorithm and the function, and also to compare the factor with the ones suggested in technical literature.

I realize that only the programmers can know the answer to my question... but I tried the same to ask!