[Insight-users] Multi-initial in the levelset processing

[???]

Hello Olivier! 
            I think we have different understanding concerning the SD.For me ,I compute the whole all pixels'SD according to the unique initial curve,and you do it by compute the points near the initial curves(narrow band?), and so, the more  the num of initial curves is ,the more easily it will converge.But the problem is if you seed one curve,the algorithm maybe  dose not converge  ,while in my implementation with the coputation of the wholeall pixels,it should converge in all conditions.But my experiment doesn't second it.The convergence are associated with the initial position ,the timestep,and the parameter.So I can not even unserstand the "take more seed curves, you have more chance  that the level set algorithm will converge faster".
            Meanwile,if there are more than two areas that you should segment,thta is  you will have to use more than one levelset equation,typically log(n) levelset functions are needed ,then how can you compute the different SD according to your closet distance computation method?
            Thank you!
			   						Yours sincerely
									Wenmin Peng
        	pwm at dsp.whu.edu.cn
         2005-06-23
-------------- next part --------------
A non-text attachment was scrubbed...
Name: face-3.gif
Type: image/gif
Size: 842 bytes
Desc: not available
Url : http://public.kitware.com/pipermail/insight-users/attachments/20050623/cb2e960c/face-3.gif