Image Feature Suggestion

Zachary Pincus zachary.pincus at yale.edu
Wed Apr 17 17:16:05 EDT 2013


Hi Ronnie,

Identifying features to track between frames is a pretty classic task in computer vision. Things that come to mind would include the simple Harris corner detector (in skimage), or more recent things like SIFT features and its variants like SURF etc. (DAISY is in skimage) might be of use. All of these tools should be in OpenCV too, IIRC. (Time to spend a while on wikipedia reading about these things and related, if it's all totally unfamiliar.)

Alternately, you could try optical flow based techniques to densely track local pixel patches between frames. 

In either case, the first thing to do would be to get a robust estimate of the global camera translation and subtract out that shake (very common operation these days... I think this is all packaged up in openCV). Then you could do the optical flow / feature tracking specifically on your spinning thing.

One basic flow might be:
(1) subtract out global camera shake
(2) find feature points to track (e.g. good Harris corners) that are not static from frame to frame (e.g. by using background estimation and subtraction)
(3) optimize an estimate of the center-of-rotaion and angle-of-rotation parameters, based on how well the feature points match between frames after applying the estimated rotation parameters.

If you provide a few sample frames, folks might be able to offer better suggestions. Over all, I think most of the tools you need for this sort of thing are in OpenCV...

Zach

On Apr 16, 2013, at 12:04 AM, Ronnie Ghose wrote:

> Hey guys,
> 
> I was wondering if you could suggest feature extraction methods to try on an image. So I was wondering given for example a gif of a wheel spinning or (something easier I should think because it has discrete branches spinning) a carousel spinning how could you identify the branches in each frame and tell the approximate radian shift, and then also do that given a slight translation in the picture - ex. a shaky video camera?
> 
> Thanks,
> Ronnie
> 
> -- 
> You received this message because you are subscribed to the Google Groups "scikit-image" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe at googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>  
>  




More information about the scikit-image mailing list