Manual Feature Extraction For Image Mosaicking/Panaroma

Hello,

I am trying to stitch multiple images of the same scene together in MATLAB in order to create one large image.

I have used this camera calibration toolbox to remove distortion from the images:(<http://www.vision.caltech.edu/bouguetj/calib_doc/>)

My problem arises when I try and 'stitch' these videos together. Whenever I use the vision toolbox examples they are never able to find enough features between each image. I have images of a checkerboard from two views, which I thought would provide a good feature to match between the images, but the matching algorithms never seem to detect them.

The example programs I have tried: http://www.mathworks.co.uk/help/vision/ug/stereo-image-rectification.html (Uses SURF features) http://www.mathworks.co.uk/help/vision/examples/video-mosaicking-1.html (Uses corner detection)

So, does anyone know of a way of manual extracting features? In a similar fashion to the cpselect function.

Thanks

Answer 1

[Anand]

I wouldn't move to manual feature extraction so soon. I don't know enough to comment on whether that is a good option.

Have you tried playing around with the parameters of the feature detectors? For example with SURF, have you tried reducing the 'MetricThreshold' parameter?

Alternatively, you can also try using a different feature detector - there are 5-6 feature detectors that you can try.

If you did want to go manual, you could write some code to select regions of interest in your images by hand and detect features only in the selected ROI.