Abstract
Effective local segmentation of contours is an important problem which arises in occluded object recognition as well as other areas. For any recognition system to perform successfully, the segmentation procedure used must be robust in presence of noise and local distortions of shape. Furthermore, it should be based on geometric invariants so that the segmentation will not be affected by arbitrary choices.
This paper proposes a new multi-scale segmentation routine for planar contours which is based on the curvature scale space representation. Curvature zero-crossing segments extracted from a continuum of scales are utilized for robust segmentation of planar curves. Curvature zero-crossing points are effectively tracked across increasing scales to ensure that the same segment is extracted only once. This approach is more robust than techniques which try to recover features/segments from a stable scale and, as a result, risk over- or under-segmentation of the input contour.
Preview
Unable to display preview. Download preview PDF.
References
N. Ayache and O. D. Faugeras. Hyper: A new approach for the recognition and positioning of 2-d objects. IEEE Trans Pattern Analysis and Machine Intelligence, 8(1):44–54, 1986.
M. Kass, A. Witkin, and D. Terzopoulos. Snakes: active contour models. In Proc International Conference on Computer Vision, pages 259–268, 1987.
A. K. Mackworth and F. Mokhtarian. Scale-based description of planar curves. In Proc Canadian Society for Computational Studies of Intelligence, pages 114–119, London, Ontario, 1984.
F. Mokhtarian. Fingerprint theorems for curvature and torsion zero-crossings. In Proc IEEE Conference on Computer Vision and Pattern Recognition, pages 269–275, San Diego, CA, 1989.
F. Mokhtarian. Zero-crossings of curvature and torsion in the limit. In Proc Asian Conference on Computer Vision, pages III:457–461, Singapore, 1995.
F. Mokhtarian. Silhouette-based object recognition with occlusion through curvature scale space. In Proc European Conference on Computer Vision, pages I:566–578, Cambridge, England, 1996.
F. Mokhtarian and A. K. Mackworth. A theory of multi-scale, curvature-based shape representation for planar curves. IEEE Trans Pattern Analysis and Machine Intelligence, 14(8):789–805, 1992.
F. Mokhtarian and S. Naito. Scale properties of curvature and torsion zerocrossings. In Proc Asian Conference on Computer Vision, pages 303–308, Osaka, Japan, 1993.
A. Rattarangsi and R. T. Chin. Scale-based detection of corners of planar curves. IEEE Trans Pattern Analysis and Machine Intelligence, 14(4):430–449, 1992.
P. L. Rosin. Representing curves at their natural scales. Pattern Recognition, 25:1315–1325, 1992.
G. Sapiro and A. Tannenbaum. Affine invariant scale-space. International Journal of Computer Vision, 11(1):25–44, 1993.
F. Stein and G. Medioni. Structural indexing: Efficient 3-d object recognition. IEEE Trans Pattern Analysis and Machine Intelligence, 14:125–145, 1992.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Mokhtarian, F. (1997). Multi-scale contour segmentation. In: ter Haar Romeny, B., Florack, L., Koenderink, J., Viergever, M. (eds) Scale-Space Theory in Computer Vision. Scale-Space 1997. Lecture Notes in Computer Science, vol 1252. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-63167-4_59
Download citation
DOI: https://doi.org/10.1007/3-540-63167-4_59
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63167-5
Online ISBN: 978-3-540-69196-9
eBook Packages: Springer Book Archive