{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

ICP - RonenGvili TheProblem Aligntwopartially...

Info iconThis preview shows pages 1–21. Sign up to view the full content.

View Full Document Right Arrow Icon
    Iterative Closest Point Ronen Gvili
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    The Problem Align two partially- overlapping meshes given initial guess for relative  transform
Background image of page 2
    Data Types Point sets Line segment sets (polylines) Implicit curves  : f(x,y,z) = 0 Parametric curves : (x(u),y(u),z(u)) Triangle sets (meshes) Implicit surfaces : s(x,y,z) = 0 Parametric surfaces (x(u,v),y(u,v),z(u,v)))
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Motivation Shape inspection Motion estimation Appearance analysis Texture Mapping Tracking
Background image of page 4
    Motivation Range images      registration
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Motivation Range images registration
Background image of page 6
    Range Scanners
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Aligning 3D Data
Background image of page 8
    Corresponding Point Set Alignment Let M be a model point set.  Let S be a scene point set. We assume : 1. N M  = N S . 2. Each point S i  correspond to M .
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Corresponding Point Set Alignment The MSE objective function : The alignment is : = = - - = - - = S S N i T i R i S N i i i S q s q R m N q f Trans s Rot m N T R f 1 2 1 2 ) ( 1 ) ( ) ( 1 ) , ( ) , ( ) , , ( S M d trans rot mse Φ =
Background image of page 10
    Aligning 3D Data If correct correspondences are known, can  find correct relative rotation/translation
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Aligning 3D Data How to find correspondences:  User input?  Feature detection?  Signatures? Alternative: assume  closest  points  correspond
Background image of page 12
    Aligning 3D Data How to find correspondences:  User input?  Feature detection?  Signatures? Alternative: assume  closest  points  correspond
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Aligning 3D Data Converges if starting position “close enough“
Background image of page 14
    Closest Point Given 2 points r 1  and r 2  , the Euclidean  distance is:  Given a point r 1  and set of points A ,  the Euclidean distance is: 2 2 1 2 2 1 2 2 1 2 1 2 1 ) ( ) ( ) ( ) , ( z z y y x x r r r r d - + - + - = - = ) , ( min ) , ( 1 .. 1 1 i n i a r d A r d =
Background image of page 15

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Finding Matches The scene   shape S is aligned to be  in the best alignment with the model  shape M. The distance of each point s of the  scene from the model is : s m d M s d M m - = min ) , (
Background image of page 16
    Finding Matches  M Y M S C Y M y y s d s m d M s d M m = = - = ) , ( ) , ( min ) , ( C – the closest point operator Y – the set of closest points to S
Background image of page 17

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    Finding Matches  Finding each match is performed in O(N M worst case. Given Y we can calculate alignment S is updated to be : ) , ( ) , , ( Y S d trans rot Φ = trans S rot S new + = ) (
Background image of page 18
    The Algorithm Init the error to  Calculate correspondence Calculate alignment Apply alignment Update error If error > threshold Y = CP(M,S),e (rot,trans,d) S`= rot(S)+trans d` = d
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
    The Algorithm function ICP(Scene,Model) begin E`   +  ; (Rot,Trans)   In Initialize-Alignment(Scene,Model); repeat       E`;     Aligned-Scene   Apply-Alignment(Scene,Rot,Trans);     Pairs   Return-Closest-Pairs(Aligned-Scene,Model);     (Rot,Trans,E`) 
Background image of page 20
Image of page 21
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}