Mirror Image

Mostly AR and Stuff

Markerless tracking with FAST

Testing outdoor markerless tracking with FAST/SURF feature detector.
The plane of the camera is not parallel to the earth, that make difficult for eye to estimate precision.

29, January, 2009 Posted by | Uncategorized | , , , , , , , , | Comments Off on Markerless tracking with FAST

FAST with SURF descriptor

Feature detected with multistage FAST and fitted with SURF descriptors
Less strict threshold give a lot more correspondences, but also some false positives

25, January, 2009 Posted by | Coding AR | , , , , , , | 44 Comments

Multiscale FAST detector

Experimenting with multiscale FAST detector with images from cell phone camera.
so far so good…

24, January, 2009 Posted by | Coding AR | , , , , , , | 2 Comments

Testin FAST feature detector

Testing FAST feature detector on the Mikolajczyk ’s dataset. Here scale space seems actually useful. With “brick wall” dataset repeatability goes form .3 to .7 with scale from 0 t 2^^3, and threshold/barrier lowering from 40 to 20.

11, January, 2009 Posted by | Coding AR | , , , , , | Comments Off on Testin FAST feature detector


I did some research on the SURF optimization. While it still possible to make it significantly faster with lazy evaluation, the problem of the scale remain. Fine-scale features are not detectedable on the bigger scale, so it doesn’t look like there is an easy way to reduce search area using only upper scale. If scale-space is not helping to reduce search area it become liability for mobile tracking – range can’t change too fast for a mobile pone, so scale of the feature will be about the same between frames.
Will try plain, not scale-space corner detectors now, starting with FAST.

5, January, 2009 Posted by | Augmented Reality, Coding AR | , , , , , , | Comments Off on From SURF to FAST

SURF scale scpace

I continue to test SURF, in respect to scale space. Scale space is essentially a pyramid of progressively more blurred or lower resolution images. The idea of scale invariant feature detection is that the “real” feature should be present at several scales – that is should be clearly detectable at several image resolution/blur levels. The interesting thing I see is, that for SURF, at least for test images from Mikolajczyk ‘s dataset, scale space seems doesn’t affect detection rate with viewpoint change. I meant that there is no difference if feature distinct in several scales or only in one. That’s actually reasonable – scale space obviously benefit detection in the blurred images, or noisy images, or repeatability/correspondence in scaled images , and “viewpoint” images form Mikolajczyk ‘s dataset are clear, high resolution and about the same scale. Nevertheless there is some possibility for optimization here.

23, December, 2008 Posted by | Coding AR | , , , , , , | Comments Off on SURF scale scpace

Testing SURF-based detectors

I have tested several modification of SURF, using original SURF Hessian, extremum of SURF-based Laplacian, Hessian-Laplace – extremum both Hessian and Laplacian and minimal eigenvalue of Hessian. They all give about the same detection rate, but original SURF Hessian give better results. Minimal eigenvalue of Hessian seems better scaling with threshold value – original Hessian absolute value could be very low, but eigenvalues are not. So this approach may have some advantage if there are potential precision loss problem, for example in fixed point calculations. A lot of high-end mobile phones still launched without hardware floating point so it still could be useful in AR or Computer Vision applications.

21, December, 2008 Posted by | Coding AR | , , , , , , , | Comments Off on Testing SURF-based detectors

Fullscreen AR Tower Defense is on the way.

Fullscreen version of the AR Tower Defense moving along nicely. I have changed default screen position from portrait to landscape. Don’t see any noticeable slowdown now. I’m thinking about increasing the number of markers from 6 to 9, but I’m not sure. Too many markers would clog the table…

PS It’s here already

4, December, 2008 Posted by | Augmented Reality, Coding AR, Symbian | , , , , , , , | Comments Off on Fullscreen AR Tower Defense is on the way.

N95 camera distortion part 2

Could be wrong.
Actually what caused my presumed mistake was this research on the Nokia N95 camera calibration
N95 Distortion Evaluation
N95 distortion map
They got huge distortion estimation – like 3% of the screen width. I don’t detect anything like this in the 320×240 viewfinder. My upper bound is like .7% (though I didn’t do complete calibration, just did some rectangles shots).

30, November, 2008 Posted by | Coding AR | , , , , , | Comments Off on N95 camera distortion part 2

N95 camera distortion ?

I’m making full screen version of my AR Tower Defense game, and stumble at some strange problem. It seems to me the camera distortion is absent if I’m getting scaled down viewfinder image (or may be it’s just cropped). If I’m getting full viewfinder image it looks like some camera distortion creep in – seems markers getting distorted. I may have to calibrate camera, and that would create portability problem. The game will work only for N95 camera…

29, November, 2008 Posted by | Coding AR, Nokia N95 | , , , , , | Comments Off on N95 camera distortion ?