Title: Regression based Bandwidth Selection for Segmentation using Parzen Windows
1Regression based Bandwidth Selection for
Segmentation using Parzen Windows
Maneesh Singh and Narendra Ahuja ECE and Beckman
Institute, Univ. of Illinois at Urbana-Champaign
2Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
3Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
4Introduction
Original
Segmented
5Introduction
- We consider segmentation of non-stationary image
signals. - Non-stationarity modeled by
which varies with location - Relationship of and
segmentation - multimodal.
- Each mode corresponds to a segment.
6Introduction
- is estimated using Parzen windows.
- Bandwidth selection critical for above estimate.
- This paper is about bandwidth selection.
- Segmentation done using Mean-Shift.
7Introduction
- Notations
- - conditional image
PDF. - - estimated cond.
image PDF (n samples, H bandwidth). - - MSE.
- - Integrated MSE
(IMSE) - - Integrated
Asymptotic MSE (IAMSE)
8Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
9Model
- Image model
- Non-stationary error PDF modeled using Parzen
Windows - modeled using a plug-in regression
estimator.
10Model
- Segmentation assumes implicit/ explicit image
model. - Our Model (flexible)
- Image signal noise
- Signal (regression) function - piecewise cont.
- Noise
- Non-stationary with unknown statistics.
- PDF smooth.
- Estimated from image data.
11Approach
- Segmentation via
- Estimation of PDF modes
- Pixel association with modes.
- We use mean-shift for above.
- Issue kernel bandwidth selection
12Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
13Bandwidth Selection
- Several Bandwidth selection schemes in
literature. - We note
- Plug-in bandwidth estimators are simplest and
most popular. - Assume an underlying data distribution (Gaussian,
locally Gaussian). - But images highly non-Gaussian.
14Bandwidth Selection
Typical 2-dim image
15Bandwidth Selection
- Histograms
- Raw values
- Wavelet residues
- Median residues
16Bandwidth Selection
- Observations
- Plug-in bandwidth estimation feasible for image
residuals (use GGD) though not for image data. - Computing residuals requires a regression
estimate - Above observation true for a wide variety of
regression functions. - Good regression-models already available
- Scale-based regression ? scale-based segmentation
17Bandwidth Selection
- Bandwidths selected to minimize the Integrated
Mean Square Error (IMSE) between the true and
estimated image PDF. - We derive expressions for AIMSE based on the
error PDF and image regression estimates.
18Bandwidth Selection
- Consistency conditions should be met
19Bandwidth Selection
20Bandwidth Selection
- Bandwidth estimated to minimize the MSE
- Asymptotic analysis required (IAMSE)
- Note we drop the conditional notation for
brevity.
21Bandwidth Selection
22Bandwidth Selection
where
- requires knowledge of and
23Bandwidth Selection
- Difficult optimization problem.
- We derive an upper bound
- Optimal bandwidths for this upper bound.
- Bound asymptotically tight.
- Detailed derivations in the paper.
24Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
25Scale Guided Segmentation
- Algorithm
- Choose scale
- Get plug-in regression estimate
- Find residuals (errors)
- Use GGD as a plug-in for error PDF
- Estimate global bandwidth parameters
- Optionally, find local BW parameters using
Abramsons Law - Use Mean-shift to find modes
26Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
27Segmentation Results
- Results using N(0,42I) smoothing function
28Segmentation Results
- Results using N(0,42I) smoothing function
29Segmentation Results
- Results using N(0,42I) smoothing function
30Segmentation Results
- Results using N(0,42I) smoothing function
31Topics
- Introduction
- Model advantages
- Bandwidth Selection
- Scale-guided segmentation
- Results
- Conclusions
32Conclusions
- Proposed a regression-based conditional PDF model
for images (Similar to Bashtannyk Hyndman,
Comp. Stats. Data Anal., 36(3), 279-298). - Regression using wavelets, median similar
results. - Noise PDF estimation using Parzen windows, GGD.
- Scale estimates easier for noise PDF.
- Better image model (than Comaniciu Meer, ITIP,
02). - Derived a Bandwidth estimation scheme.
- Leads to (spatial) scale-based segmentation
framework.
33The Object Selection Problem
- Interactively extract objects embedded in images
and video for compositing and manipulation - Application areas special effects, graphic
design, TV broadcast - Key to enabling intelligent editing operators
34Related Work
- Vector graphic editors
- GRANDMA CMU
- PerSketch Xerox PARC
- Edge-based selection
- Intelligent Scissors BYU, Adobe
- Active Contours Various
- Alpha channel-based
- Commercial Tools - Photoshop, Ultimatte, KnockOut
- Alpha Estimation - Ruzon-Tomasi, Chuang et al.
35Segmentation-based Selection with Freehand
Sketches
36Rose Example Selection Results
Photoshop
Segmentation-based
It can be seen that the segmentation-based tool
produced higher-quality object boundaries
37Limitation of Segmentation and Edge-based
Representation
- Diffused Edges are present in most images due to
- Optical defocus (intentional or otherwise)
- Motion blur
- Finite Sensor Resolution
- Antialiasing (synthetic images)
- Compression
- Texture boundary
- Need an alpha channel to represent diffused edges
and recover foreground and background
38Boundary Region Decomposition ICCV01
- Use simple segmentation algorithm that decomposes
image into approximately convex segment - Use the Delaunay triangulation of the segment
centroids as decomposition
Lines can now be matched to centroids using the
triangles.
39Local Boundary Analysis
- Within each triangle
- Find linear approximation of boundary using
initial selection - Finer-scale segmentation
- Classify new segment centroids using discriminant
- Extract pixel samples in a window around centroid
- Compute alpha channel
40Test Images
41Selection with Points
Click around and select
42Selection with Loops
43Selection with Loops
44A Better Triangulation CVPR01
- Want triangles to satisfy the following
properties - Vertices lie on medial axis of segments
- Each triangle is completely contained in at most
three regions - For an edge between two vertices, one in region A
and one in region B, the edge must lie entirely
in the union of A and B.
1. Cover junctions
2. Cover openings
45Selection with New Triangulation
Details
46Contact
- msingh_at_uiuc.edu , n-ahuja_at_uiuc.edu