Hi,
I've been playing with around with OpenCV, and have been trying to implement a simple logo detector.
I started with the feature homography [tutorial](http://docs.opencv.org/doc/tutorials/features2d/feature_homography/feature_homography.html). The idea was to use the same workflow to detect a logo of interest.
[Here is the relevant portion of my code](https://gist.github.com/anonymous/5606da4e3e37380ed5e5).
Here is an [output image](http://i.imgur.com/XEObrUC.png). Focus on the area in and around the ROGERS logo. You'll see the matching keypoints drawn inside the ROGERS logo, but my attempt to draw a box around the detected logo is not exactly successful. Other logos and non-logo objects all seem to suffer from the same issue, and I can't see what it is.
Can anyone see anything fundamentally wrong in my code, or suggest how I might get better bounding box locations?
Any advice/suggestions is much appreciated.
Thanks in advance!
↧