Download presentation
Presentation is loading. Please wait.
1
Fast, Multiscale Image Segmentation: From Pixels to Semantics Ronen Basri The Weizmann Institute of Science Joint work with Achi Brandt, Meirav Galun, Eitan Sharon
4
Camouflage
5
Camouflage Malik et al.’s “Normalized cuts”
11
Our Results
12
Segmentation by Weighted Aggregation A multiscale algorithm: Optimizes a global measure Returns a full hierarchy of segments Linear complexity Combines multiscale measurements: –Texture –Boundary integrity
13
The Pixel Graph Couplings (weights) reflect intensity similarity Low contrast – strong coupling High contrast – weak coupling
14
Normalized-cut Measure Minimize:
15
Saliency Measure Minimize:
16
Multiscale Computation of Ncuts Our objective is to rapidly find the segments (0-1 partitions) that optimize For single-node cuts we simply evaluate For multiple-node cuts we perform “soft contraction” using coarsening procedures from algebraic multigrid solvers of PDEs.
17
Coarsening the Graph Suppose we can define a sparse mapping such that for all minimal states
18
Coarse Energy Then P T WP, P T LP define a new (smaller) graph
19
Recursive Coarsening
20
Representative subset
21
Recursive Coarsening For a salient segment :, sparse interpolation matrix
22
Weighted Aggregation aggregate
23
Hierarchical Graph Pyramid of graphs Soft relations between levels Segments emerge as salient nodes at some level of the pyramid
24
Importance of Soft Relations
25
Physical Motivation Our algorithm is motivated by algebraic multigrid solutions to heat or electric networks u - temperature/potential a(x, y) – conductivity At steady state largest temperature differences are along the cuts AMG coarsening is independent of f
26
Determine the Boundaries 1 0 0 1,0,0,…,0 P
27
Hierarchy in SWA
28
Texture Examples
29
Filters (From Malik and Perona) Oriented filters Center- surround
30
A Chicken and Egg Problem Problem: Coarse measurements mix neighboring statistics Solution: Support of measurements is determined as the segmentation process proceeds Hey, I was here first
31
Texture Aggregation Aggregates assumed to capture texture elements Compare neighboring aggregates according to the following statistics: –Multiscale brightness measures –Multiscale shape measures –Filter responses Use statistics to modify couplings
32
Recursive Computation of Measures Given some measure of aggregates at a certain level (e.g., orientation) At every coarser level we take a weighted sum of this measure from previous level The result can be used to compute the average, variance or histogram of the measure Complexity is linear
33
Use Averages to Modify the Graph
34
Adaptive vs. Rigid Measurements Averaging Our algorithm - SWA Original Geometric
35
Adaptive vs. Rigid Measurements Interpolation Geometric Original Our algorithm - SWA
36
Adaptive vs. Rigid Measurements
41
Texture Aggregation Fine (homogeneous) Coarse (heterogeneous)
42
Multiscale Variance Vector
44
Variance: Avoid Mixing aggregationSliding window
45
Leopard
46
More Leopards…
47
And More…
48
Birds
49
More Animals
50
Boat
51
Malik’s Ncuts
52
Key Differences Optimize a global measure (like Malik’s Ncuts) Hierarchy with soft relations (unlike agglomerative/graph contraction) Combine texture measurements while avoiding the “chicken and egg problem”
53
Complexity Every level contains about half the nodes of the previous level: Total #nodes 2 X #pixels All connections are local, cleaning small weights Top-down sharpening: constant number of levels Linear complexity Implementation: 5 seconds for 400x400
54
Average intensity Texture Shape Representation
55
Relevance to Biology? Layers of retinotopic maps Feed-forward (fine-to-coarse) progression Feedback refinement Edges determine both boundaries and textures Segmentation determined by combinations of cues Coarse nodes represent abstract shapes and properties
56
Matching (with Chen Brestel)
57
More…
58
Data: Filippi 30 slices, 180x220 in 3 minutes MRI Data
59
MS Lesion Detection Tagged Our results Data: Filippi
60
Tagged Our results
61
Data: Filippi Tagged Our results
62
Data: Filippi Tagged Our results
63
2D Segmentation Data: Filippi
64
3D Segmentation
65
Cell Movement
66
Summary image segments Shape properties Leopard
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.