Affine Resilient Curvature Scale-Space Corner Detector

Apply the affine-length parameterized curvature to the CSS corner detection technique.

Description

1. Find edge image using the Canny edge detector.
2. Extract edges from the edge image:
   a. fill gaps if they are within a range and select long edges,
   b. find T-junctions and mark them as T-corners.
3. Parameterize each edge with its affine-length.
4. For each parameterized edge, compute absolute curvature at an appropriate scale in {σm, σm+1, σm+2} and determine corners by comparing the curvature maxima to the corresponding curvature threshold of edge in {tm, tm+1, tm+2} and the neighboring minima.
5. Track the corners down to the lowest scale considering a small
neighborhood in order to improve localization.
6. Further track the corners on the original edge at the lowest scale considering the same neighborhood size.
7. Remove multiple occurrences of same corners, compare Tcorners with the tracked corners and add those T-corners which are far away from the detected corners.

Acknowledgements

Robust Image Corner Detection Based On The Chord To Point Distance Accumulation Technique and A Fast Corner Detector Based On The Chord To Point Distance Accumulation Technique inspired this file.

• MATLAB release: MATLAB 7 (R14)
• Other requirements: (Should be compatible with Unix as well as Windows)