questions about "surf.surf" function

[Antonis Polykratis]

Hello,

I have two questions

1) I try to understand the following line of code that I saw in a text:
surf.surf(image)[:, 5:]
Is there any known (general) reason for the above slicing? Why not surf.surf(image)?

2) surf.surf(image) is returning empty result in some photos. I think that these photos are small.
I would expect that in these cases the features would be less than normal but not zero. (So in these
cases I have to add logic to exclude those images).
Any comment on that?

Thanks for your time.
Antonis

[Luis Pedro Coelho]

Hi Antonis,

> 1) I try to understand the following line of code that I saw in a text:
> surf.surf(image)[:, 5:]
> Is there any known (general) reason for the above slicing? Why not
> surf.surf(image)?

The first few elements contain the position (y,x), the scale, the score,
the Laplacian. These are ignored by the code above.

See:
http://mahotas.readthedocs.org/en/latest/surf.html

> 2) surf.surf(image) is returning empty result in some photos. I think
> that
> these photos are small.
> I would expect that in these cases the features would be less than normal
> but not zero. (So in these
> cases I have to add logic to exclude those images).
> Any comment on that?

This probably means that no interest points were detected. Perhaps, try
dense sampling:

see
http://mahotas.readthedocs.org/en/latest/surf.html?highlight=dense#api-documentation

HTH
Luis

[Antonis Polykratis]

Thank you Luis!

A.