TY - GEN
T1 - Deformable markers
T2 - Image Algebra and Morphological Image Processing II
AU - Rougon, Nicolas F.
AU - Preteux, Francoise
PY - 1991/1/1
Y1 - 1991/1/1
N2 - The concept of a deformable (active) marker of an image, defined as a numerical marker whose support is limited by a deformable (active) contour, is introduced. An active marker is specified by defining (1) an interface deformable model as well as the deformation process associated with it, and (2) an interaction process between the marker and the external field. We present a particular active marker model relying on a novel expanding-contracting inhomogeneous membrane/thin-plate model called a g-snake, and an interaction process based on controlled morphological marking techniques. We show that by allowing the external force field to be simplified and taking into account a global source of information, active markers provide a consistent solution to three major problems encountered by active contour models when applied to the segmentation of highly noisy images: (1) sensitivity to initialization, (2) undesirable attractions by non-significant localized or regionalized zones in the image, and (3) no relationship between their state of equilibrium and the real contours to be extracted. The efficiency and robustness of the method are demonstrated on ultrasound medical images.
AB - The concept of a deformable (active) marker of an image, defined as a numerical marker whose support is limited by a deformable (active) contour, is introduced. An active marker is specified by defining (1) an interface deformable model as well as the deformation process associated with it, and (2) an interaction process between the marker and the external field. We present a particular active marker model relying on a novel expanding-contracting inhomogeneous membrane/thin-plate model called a g-snake, and an interaction process based on controlled morphological marking techniques. We show that by allowing the external force field to be simplified and taking into account a global source of information, active markers provide a consistent solution to three major problems encountered by active contour models when applied to the segmentation of highly noisy images: (1) sensitivity to initialization, (2) undesirable attractions by non-significant localized or regionalized zones in the image, and (3) no relationship between their state of equilibrium and the real contours to be extracted. The efficiency and robustness of the method are demonstrated on ultrasound medical images.
UR - https://www.scopus.com/pages/publications/0025813644
M3 - Conference contribution
AN - SCOPUS:0025813644
SN - 0819406961
T3 - Proceedings of SPIE - The International Society for Optical Engineering
SP - 78
EP - 89
BT - Proceedings of SPIE - The International Society for Optical Engineering
PB - Publ by Int Soc for Optical Engineering
Y2 - 23 July 1991 through 24 July 1991
ER -