Segmentation with non-linear constraints on appearance, complexity, and geometry Yuri Boykov Western Univesity Andrew Delong Olga Veksler Lena Gorelick.

Slides:



Advertisements
Similar presentations
POSE–CUT Simultaneous Segmentation and 3D Pose Estimation of Humans using Dynamic Graph Cuts Mathieu Bray Pushmeet Kohli Philip H.S. Torr Department of.
Advertisements

O BJ C UT M. Pawan Kumar Philip Torr Andrew Zisserman UNIVERSITY OF OXFORD.
MRI Brain Extraction using a Graph Cut based Active Contour Model Noha Youssry El-Zehiry Noha Youssry El-Zehiry and Adel S. Elmaghraby Computer Engineering.
SOFT SCISSORS: AN INTERACTIVE TOOL FOR REALTIME HIGH QUALITY MATTING International Conference on Computer Graphics and Interactive Techniques ACM SIGGRAPH.
Graph cut Chien-chi Chen.
Graph Cut Algorithms for Computer Vision & Medical Imaging Ramin Zabih Computer Science & Radiology Cornell University Joint work with Y. Boykov, V. Kolmogorov,
Lena Gorelick Joint work with Frank Schmidt and Yuri Boykov Rochester Institute of Technology, Center of Imaging Science January 2013 TexPoint fonts used.
Interactive Segmentation with Super-Labels Andrew Delong Western Yuri BoykovOlga VekslerLena GorelickFrank Schmidt TexPoint fonts used in EMF. Read the.
Stereo Many slides adapted from Steve Seitz. Binocular stereo Given a calibrated binocular stereo pair, fuse it to produce a depth image Where does the.
I Images as graphs Fully-connected graph – node for every pixel – link between every pair of pixels, p,q – similarity w ij for each link j w ij c Source:
The University of Ontario CS 4487/9687 Algorithms for Image Analysis Multi-Label Image Analysis Problems.
ICCV 2007 tutorial on Discrete Optimization Methods in Computer Vision part I Basic overview of graph cuts.
C. Olsson Higher-order and/or non-submodular optimization: Yuri Boykov jointly with Western University Canada O. Veksler Andrew Delong L. Gorelick C. NieuwenhuisE.
GrabCut Interactive Image (and Stereo) Segmentation Carsten Rother Vladimir Kolmogorov Andrew Blake Antonio Criminisi Geoffrey Cross [based on Siggraph.
Stephen J. Guy 1. Photomontage Photomontage GrabCut – Interactive Foreground Extraction 1.
1 s-t Graph Cuts for Binary Energy Minimization  Now that we have an energy function, the big question is how do we minimize it? n Exhaustive search is.
Graph-based image segmentation Václav Hlaváč Czech Technical University in Prague Faculty of Electrical Engineering Department of Cybernetics Prague, Czech.
GrabCut Interactive Image (and Stereo) Segmentation Joon Jae Lee Keimyung University Welcome. I will present Grabcut – an Interactive tool for foreground.
Learning with Inference for Discrete Graphical Models Nikos Komodakis Pawan Kumar Nikos Paragios Ramin Zabih (presenter)
Pseudo-Bound Optimization for Binary Energies Presenter: Meng Tang Joint work with Ismail Ben AyedYuri Boykov 1 / 27.
Optimal solution of binary problems Much material taken from :  Olga Veksler, University of Western Ontario
Learning to Detect A Salient Object Reporter: 鄭綱 (3/2)
1 Can this be generalized?  NP-hard for Potts model [K/BVZ 01]  Two main approaches 1. Exact solution [Ishikawa 03] Large graph, convex V (arbitrary.
Robust Higher Order Potentials For Enforcing Label Consistency
Schedule Introduction Models: small cliques and special potentials Tea break Inference: Relaxation techniques:
Today: Image Segmentation Image Segmentation Techniques Snakes Scissors Graph Cuts Mean Shift Wednesday (2/28) Texture analysis and synthesis Multiple.
2010/5/171 Overview of graph cuts. 2010/5/172 Outline Introduction S-t Graph cuts Extension to multi-label problems Compare simulated annealing and alpha-
Stereo & Iterative Graph-Cuts Alex Rav-Acha Vision Course Hebrew University.
Graph Cut based Inference with Co-occurrence Statistics Ľubor Ladický, Chris Russell, Pushmeet Kohli, Philip Torr.
Stereo Computation using Iterative Graph-Cuts
What Energy Functions Can be Minimized Using Graph Cuts? Shai Bagon Advanced Topics in Computer Vision June 2010.
Measuring Uncertainty in Graph Cut Solutions Pushmeet Kohli Philip H.S. Torr Department of Computing Oxford Brookes University.
Perceptual Organization: Segmentation and Optical Flow.
Automatic User Interaction Correction via Multi-label Graph-cuts Antonio Hernández-Vela, Carlos Primo and Sergio Escalera Workshop on Human Interaction.
Graph-based Segmentation
MRFs and Segmentation with Graph Cuts Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem 03/31/15.
Mutual Information-based Stereo Matching Combined with SIFT Descriptor in Log-chromaticity Color Space Yong Seok Heo, Kyoung Mu Lee, and Sang Uk Lee.
Graph Cut & Energy Minimization
MRFs and Segmentation with Graph Cuts Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem 02/24/10.
Graph Cut Algorithms for Binocular Stereo with Occlusions
Interactive Graph Cuts for Optimal Boundary & Region Segmentation of Objects in N-D Images (Fri) Young Ki Baik, Computer Vision Lab.
Relative Volume Constraints for 3D Image Editing Computer Vision Group TU Munich Eno Töppe, Claudia Nieuwenhuis, Daniel Cremers May 25th, 2012.
Object Stereo- Joint Stereo Matching and Object Segmentation Computer Vision and Pattern Recognition (CVPR), 2011 IEEE Conference on Michael Bleyer Vienna.
Lena Gorelick joint work with O. Veksler I. Ben Ayed A. Delong Y. Boykov.
Stereo Many slides adapted from Steve Seitz.
Stereo Many slides adapted from Steve Seitz. Binocular stereo Given a calibrated binocular stereo pair, fuse it to produce a depth image image 1image.
Auxiliary Cuts for General Classes of Higher-Order Functionals 1 Ismail Ben Ayed, Lena Gorelick and Yuri Boykov.
O BJ C UT M. Pawan Kumar Philip Torr Andrew Zisserman UNIVERSITY OF OXFORD.
The University of Ontario CS 4487/9687 Algorithms for Image Analysis Multi-Label Image Analysis Problems.
Gaussian Mixture Models and Expectation-Maximization Algorithm.
Machine Learning – Lecture 15
Yuri Boykov jointly with
Pseudo-Bound Optimization for Binary Energies Meng Tang 1 Ismail Ben Ayed 2 Yuri Boykov 1 1 University of Western Ontario, Canada 2 GE Healthcare Canada.
CS654: Digital Image Analysis Lecture 28: Advanced topics in Image Segmentation Image courtesy: IEEE, IJCV.
A global approach Finding correspondence between a pair of epipolar lines for all pixels simultaneously Local method: no guarantee we will have one to.
Photoconsistency constraint C2 q C1 p l = 2 l = 3 Depth labels If this 3D point is visible in both cameras, pixels p and q should have similar intensities.
Image segmentation.
Energy minimization Another global approach to improve quality of correspondences Assumption: disparities vary (mostly) smoothly Minimize energy function:
MRFs and Segmentation with Graph Cuts Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem 03/27/12.
Graph Cuts vs. Level Sets
GrabCut Interactive Foreground Extraction using Iterated Graph Cuts Carsten Rother Vladimir Kolmogorov Andrew Blake Microsoft Research Cambridge-UK.
Advanced Appearance and Shape Models
Markov Random Fields with Efficient Approximations
Nonparametric Semantic Segmentation
Secrets of GrabCut and Kernel K-means
Multiway Cut for Stereo and Motion with Slanted Surfaces
Efficient Graph Cut Optimization for Full CRFs with Quantized Edges
Multi-view reconstruction
Discrete Optimization Methods Basic overview of graph cuts
“Traditional” image segmentation
Presentation transcript:

Segmentation with non-linear constraints on appearance, complexity, and geometry Yuri Boykov Western Univesity Andrew Delong Olga Veksler Lena Gorelick Frank Schmidt Anton Osokin Hossam Isack IPAM February 2013

Overview Standard linear constraints on segments – intensity log-likelihoods, volumetric ballooning, etc. 1. Basic non-linear regional constraints – enforcing intensity distribution (KL, Bhattacharia, L 2 ) – constraints on volume and shape 2. Complexity constraints (label costs) – unsupervised and supervised image segmentation, compression – geometric model fitting 3. Geometric constraints – unsupervised and supervised image segmentation, compression – geometric model fitting (lines, circles, planes, homographies, motion,…)

Image segmentation Basics 3 S

Linear appearance of region S Examples of potential functions f Log-likelihoods Chan-Vese Ballooning

Part 1 Basic non-linear regional functionals

Standard Segmentation Energy 6 Fg Bg Intensity Probability Distribution Target Appearance Resulting Appearance

Minimize Distance to Target Appearance Model 7 Non-linear harder to optimize regional term Non-linear harder to optimize regional term

– appearance models – shape non-linear regional term 8

Related Work Can be optimized with gradient descent – first order approximation 9 Ben Ayed et al. Image Processing 2008, Foulonneau et al., PAMI 2006 Foulonneau et al., IJCV 2009 Ben Ayed et al. Image Processing 2008, Foulonneau et al., PAMI 2006 Foulonneau et al., IJCV 2009 We use higher-order approximation based on trust region approach

a general class of non-linear regional functionals

Regional Functional Examples Volume Constraint 11

Bin Count Constraint 12 Regional Functional Examples

Histogram Constraint 13

Regional Functional Examples Histogram Constraint 14

Regional Functional Examples Histogram Constraint 15

Shape Prior Volume Constraint is a very crude shape prior Can be generalized to constraints for a set of shape moments m pq 16

Volume Constraint is a very crude shape prior Shape Prior 17

Shape Prior using Shape Moments m pq 18

Shape Prior Constraint Shape Prior using Shape moments 19 Dist(),

Optimization of Energies with Higher-order Regional Functionals 20

Gradient Descent (e.g. level sets) Gradient Descent First Order Taylor Approximation for R(S) First Order approximation for B(S) (“curvature flow”) Robust with tiny steps – Slow – Sensitive to initialization 21 Ben Ayed et al. CVPR 2010, Freedman et al. tPAMI 2004 Ben Ayed et al. CVPR 2010, Freedman et al. tPAMI 2004

Energy Specific vs. General Speedup via energy- specific methods – Bhattacharyya Distance – Volume Constraint We propose – trust region optimization algorithm for general high-order energies – higher-order (non-linear) approximation Ben Ayed et al. CVPR 2010, Werner, CVPR2008 Woodford, ICCV2009 Ben Ayed et al. CVPR 2010, Werner, CVPR2008 Woodford, ICCV

General Trust Region Approach An overview The goal is to optimize Trust region Trust Region Sub-Problem First Order Taylor for R(S) Keep quadratic B(S) First Order Taylor for R(S) Keep quadratic B(S) 23

General Trust Region Approach An overview The goal is to optimize Trust Region Sub-Problem 24

Solving Trust Region Sub-Problem Constrained optimization minimize Unconstrained Lagrangian Formulation minimize Can be optimized globally u using graph-cut or convex continuous formulation L 2 distance can be approximated with unary terms [Boykov, Kolmogorov, Cremers, Delong, ECCV’06] L 2 distance can be approximated with unary terms [Boykov, Kolmogorov, Cremers, Delong, ECCV’06] 25

Approximating distance [BKCD – ECCV 2006]

Trust Region Standard (adaptive) Trust Region – Control of step size d Lagrangian Formulation – Control of the Lagrange multiplier λ 27 λ

Spectrum of Solutions for different λ or d 28 Newton step “Gradient Descent” Exact Line Search (ECCV12) Newton step “Gradient Descent” Exact Line Search (ECCV12)

Volume Constraint for Vertebrae segmentation Log-Lik. + length + volume Fast Trust Region Initializations Log-Lik. + length 29

Appearance model with KL Divergence Constraint Init Fast Trust Region “Gradient Descent” Exact Line Search 30 Appearance model is obtained from the ground truth

Appearance Model with Bhattacharyya Distance Constraint 31 “ “ Fast Trust Region “Gradient Descent” Exact Line Search Appearance model is obtained from the ground truth

Shape prior with Tchebyshev moments for spine segmentation 32 Log-Lik. + length + Shape Prior Fast Trust Region Second order Tchebyshev moments computed for the user scribble

Part 2 Complexity constraints on appearance

Segment appearance ? sum of log-likelihoods (linear appearance) – histograms – mixture models assumes i.i.d. pixels with constraints – based on information theory (MDL complexity) – based on geometry (anatomy, scene layout) now: allow sub-regions (n-labels)

Natural Images: GMM or MRF? 35 are pixels in this image i.i.d.?NO!

Natural Images: GMM or MRF? 36

Natural Images: GMM or MRF? 37

Natural Images: GMM or MRF? 38

Binary graph cuts 39 [Boykov & Jolly, ICCV 2001] [Rother, Kolmogorov, Blake, SIGGRAPH 2004]

40 [Boykov & Jolly, ICCV 2001] [Rother, Kolmogorov, Blake, SIGGRAPH 2004] Binary graph cuts

41 [Boykov & Jolly, ICCV 2001] [Rother, Kolmogorov, Blake, SIGGRAPH 2004] Binary graph cuts

Objects within image can be as complex as image itself Where do we draw the line? A Spectrum of Complexity 42 MRF?GMM? Gaussian? object recognition??

Single Model Per Class Label 43 Pixels are identically distributed inside each segment

Multiple Models Per Class Label 44 Now pixels are not identically distributed inside each segment

Multiple Models Per Class Label 45 Now pixels are not identically distributed inside each segment Our Energy ≈ Supervised Zhu & Yuille!

46 boundary length MDL regularizer + color similarity +  -expansion (graph cuts) can handle such energies with some optimality guarantees [IJCV’12,CVPR’10] Leclerc, PAMI’92 Zhu & Yuille. PAMI’96; Tu & Zhu. PAMI’02 Unsupervised clustering of pixels

(unsupervised image segmentation) Fitting color models information theory (MDL) interpretation: = number of bits to compress image I losslessly each label L represents some distribution Pr(I|L)

(unsupervised image segmentation) Fitting color models Label costs only Delong, Osokin, Isack, Boykov, IJCV 12 (UFL-approach)

(unsupervised image segmentation) Fitting color models Spatial smoothness only [Zabih & Kolmogorov, CVPR 04]

(unsupervised image segmentation) Fitting color models Spatial smoothness + label costs Zhu & Yuille, PAMI 1996 (gradient descent) Delong, Osokin, Isack, Boykov, IJCV 12 (a-expansion)

Fitting planes (homographies) Delong, Osokin, Isack, Boykov, IJCV 12 (a-expansion)

Fitting planes (homographies) Delong, Osokin, Isack, Boykov, IJCV 12 (a-expansion)

Back to interactive segmentation 53

54 segmentationcolour models“sub-labeling” EMMCVPR 2011

Main Idea Standard MRF: Two-level MRF: 55 object MRF GMMs background MRF image-level MRF object GMMbackground GMM image-level MRF unknown number of labels in each group!

Our multi-label energy functional 56 Penalizes number of GMMs (labels) – prefer fewer, simpler models – MDL / information criterion regularize “unsupervised” aspect Discontinuity cost c is higher between labels of different categories data costssmooth costs label costs GMMs set of labels L background labels object labels two categories of labels (respecting hard-constraints)

More Examples 57 standard 1-level MRF 2-level MRF

More Examples 58 2-level MRF standard 1-level MRF

Beyond GMMs 59 GMMs GMMs + planes plane GMMs only

Part 3 Geometric constraints on sub-segments

Sub-labels maybe known apriori -known parts of organs or cells -interactivity becomes optional Geometric constraints should be added - human anatomy (medical imaging) - or known scene layout (computer vision) Towards biomedical image segmentation…

Illustrative Example We want to distinguish between these objects 62 input what mixed model giveswhat we want mixed appearance model:

main ideas – sub-labels with distinct appearance models (as earlier) – basic geometric constraints between parts -inclusion/exclusion -expected distances and margins For some constraints globally optimal segmentation can be computed in polynomial time 63

Illustrative Example Model as two parts: “light X contains dark Y” 64 Center object the only object that fits two-part model (no localization)

‘layer cake’ a-la Ishikawa’03 Our Energy 65 variables Let over objects L and pixels P

Surface Regularization Terms Standard regularization of each independent surface 66 Let over objects L and pixels P

Geometric Interaction Terms Inter-surface interaction 67 Let over objects L and pixels P

So What Can We Do? Nestedness/inclusion of sub-segments ICCV 2009 (submodular, exact solution) Spring-like repulsion of surfaces, minimum distance ICCV 2009 (submodular, exact solution) Spring-like attraction of surfaces, Hausdorf distance ECCV 2012 (approximation) Extends Li, Wu, Chen & Sonka [PAMI’06] – no pre-computed medial axes – no topology constraints 68

Applications Medical Segmentation – Lots of complex shapes with priors between boundaries – Better domain-specific models Scene Layout Estimation – Basically just regularize Hoiem-style data terms [4] 69

Application: Medical 70 our result input

Application: Medical 71 full body MRI two-part model

Application: Medical 72 full body MRI

Conclusions linear functionals are very limited – be careful with i.i.d. (histograms or mixture models) higher-order regional functionals use sub-segments regularized by – complexity or sparcity (MDL, information theory) – geometry (based on anatomy) literature: ICCV’09, EMMCVPR’11, ICCV’11,, IJCV12, ECCV’12