Title :
Accurate segmentation of image with large gaps by Geometric Active Contours
Author :
Lu, Hongyu ; Duan, Chaijie ; Bao, Shanglian
Author_Institution :
Beijing City Key Lab. of Med. Phys. & Eng., Peking Univ., Peking
Abstract :
Geometric active contours (GAC) suffers from the large edge gap problem seriously. Further, high order numerical approach is required to bring the image force into effect with level set method. A first order numerical scheme, which can add the image force into the GAC equation effectively, is proposed to solve these two difficulties. This technique is accomplished by simulating the wave propagating in the isotropic substance, instead of the anisotropic substance represented in many works. The function of the image force is activated by moving the contour forward in the same speed without changing its original direction till it reaches the object boundaries. In addition, a general image force model is introduced to support the proposed scheme. Those approaches can give satisfied solutions to the large gap problem, and significantly improve the segmentation accuracy indicated by the vector field of the image force.
Keywords :
computational geometry; edge detection; image segmentation; geometric active contour; image force model; image segmentation; isotropic substance; level set method; Active contours; Anisotropic magnetoresistance; Biomedical imaging; Cities and towns; Equations; Image segmentation; Laboratories; Level set; Physics; Shape;
Conference_Titel :
Signal Processing, 2008. ICSP 2008. 9th International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4244-2178-7
Electronic_ISBN :
978-1-4244-2179-4
DOI :
10.1109/ICOSP.2008.4697354