DocumentCode
2182214
Title
Intelligent segmentation tools
Author
Barrett, William A. ; Reese, L. Jack ; Mortensen, Eric N.
Author_Institution
Brigham Young Univ., Provo, UT, USA
fYear
2002
fDate
2002
Firstpage
217
Lastpage
220
Abstract
Intelligent Scissors and Intelligent Paint are complementary interactive image segmentation tools that allow a user to quickly and accurately select objects of interest using simple gesture motions with a mouse. With Intelligent Scissors, when the cursor position comes in proximity to an object edge, a live-wire boundary "snaps" to, and wraps around the object of interest. The Intelligent Paint tool uses the cursor position to sample the image data interior to the object and grows the current region, in discrete, snapping increments, to include similar neighboring regions. Both techniques make use of a watershed algorithm called toboganning. With Intelligent Scissors, image segmentation is formulated as a piece-wise globally optimal graph searching problem, while Intelligent Paint approaches it as an adaptive, cost-ordered connected component labelling scheme. Using these tools, objects or regions can be selected in a few seconds, with better accuracy and reproducibility than can be obtained using manual selection tools. In particular interobserver reproducibility using intelligent segmentation tools is far better than intraobserver reproducibility using manual segmentation methods.
Keywords
artificial intelligence; edge detection; graphs; image segmentation; medical image processing; software tools; Intelligent Paint; Intelligent Scissors; adaptive cost-ordered connected component labelling scheme; intraobserver reproducibility; manual segmentation methods; objects of interest; piece-wise globally optimal graph searching problem; quick accurate selection; snapping increments; toboganning; watershed algorithm; Active contours; Competitive intelligence; Data mining; Feedback; Image segmentation; Labeling; Lifting equipment; Mice; Paints; Robustness;
fLanguage
English
Publisher
ieee
Conference_Titel
Biomedical Imaging, 2002. Proceedings. 2002 IEEE International Symposium on
Print_ISBN
0-7803-7584-X
Type
conf
DOI
10.1109/ISBI.2002.1029232
Filename
1029232
Link To Document