12.07.2015 Views

Metrics of curves in shape optimization and analysis - Andrea Carlo ...

Metrics of curves in shape optimization and analysis - Andrea Carlo ...

Metrics of curves in shape optimization and analysis - Andrea Carlo ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

10.1 Sobolev-type metricsRecently <strong>in</strong> [53–58, 35] Yezzi–M.–Sundaramoorthi studied a family <strong>of</strong> Sobolevtypemetrics. Let D s := 1|c ′ | ∂ θ be the derivative with respect to arc parameter.Let j ≥ 1 be an <strong>in</strong>teger 14 <strong>and</strong>∫〈h 1 , h 2 〉 Hj := 〈Dsh j 1 , Dsh j 2 〉 ds (10.1)0cwhere ∫ · · · ds was def<strong>in</strong>ed <strong>in</strong> 1.9. Let λ > 0 a fixed constant; we def<strong>in</strong>e thecSobolev-type metrics∫〈h 1 , h 2 〉 H j := 〈h 1 , h 2 〉 ds + λL 2j 〈h 1 , h 2 〉 Hj(10.2)〈h 1 , h 2 〉 ˜Hj:=c〈 ∫ h 1 ds,c∫c〉h 2 ds + λL 2j 〈h 1 , h 2 〉 Hj0where L = len(c). Notice that these metrics are geometric:0(10.3)• they are easily seen to be <strong>in</strong>variant w.r.to rotations <strong>and</strong> translations, (<strong>in</strong> astronger sense than <strong>in</strong> page 39, <strong>in</strong>deed <strong>in</strong> this case〈h 1 , h 2 〉 Ac = 〈h 1 , h 2 〉 c , 〈Rh 1 , Rh 2 〉 c = 〈h 1 , h 2 〉 cfor any Euclidean transformation A <strong>and</strong> rotation R);• they are reparameterization <strong>in</strong>variant due to the usage <strong>of</strong> the arc parameter<strong>in</strong> derivatives <strong>and</strong> <strong>in</strong>tegrals;• they are scale <strong>in</strong>variant, s<strong>in</strong>ce the normaliz<strong>in</strong>g factors L 2j make them0-homogeneous w.r.to rescal<strong>in</strong>g <strong>of</strong> the curve.For this last reason, we redef<strong>in</strong>e the H 0 metric to be∫〈h 1 , h 2〉H := h 1 · h 2 ds (10.4)0so that it is aga<strong>in</strong> 0-homogeneous.This is a conformal version <strong>of</strong> the H 0 metricdef<strong>in</strong>ed <strong>in</strong> eqn. (2.4), so a gradient descent flow is a time reparameterization <strong>of</strong>the flow for the orig<strong>in</strong>al H 0 metric; <strong>and</strong> the <strong>in</strong>duced distance is aga<strong>in</strong> degenerate.When we will present a <strong>shape</strong> <strong>optimization</strong> energy E <strong>and</strong> we will m<strong>in</strong>imizeE us<strong>in</strong>g a gradient descent flow driven by the H j or ˜H j gradient <strong>of</strong> E, we willcall the result<strong>in</strong>g algorithm a Sobolev active contour method (abbreviatedas SAC <strong>in</strong> the follow<strong>in</strong>g).14 It is though possible to def<strong>in</strong>e Sobolev metrics for any j ∈ lR, j > 0; see Prop. 3.1 <strong>in</strong> [57].c61

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!