Ch 10 Image Segmentation

Ideally, partition an image into

regions corresponding to real
world objects.

CSE 803 Fall 2008 Stockman 1

Goals of segmentation

CSE 803 Fall 2008 Stockman 2

Segments formed by K-means

CSE 803 Fall 2008 Stockman 3

Segmentation attempted via
contour/boundary detection

CSE 803 Fall 2008 Stockman 4

Clustering versus region-growing

CSE 803 Fall 2008 Stockman 5

Clustering versus region-growing

CSE 803 Fall 2008 Stockman 6

K-means clustering as before:
vectors can contain color+texture

CSE 803 Fall 2008 Stockman 7

K-means

CSE 803 Fall 2008 Stockman 8

Histograms can show modes

CSE 803 Fall 2008 Stockman 9

Otsu’s method
assumes K=2. It
searches for the
threshold t that
optimizes the intra
class variance.

CSE 803 Fall 2008 Stockman 10

Ohlander
bifurcated
the
histogram
recursively

CSE 803 Fall 2008 Stockman 11

Recursive histogram-controlled
segmentation

CSE 803 Fall 2008 Stockman 12

URL’s of other work
 Segmentation lecture
http://robots.stanford.edu/cs223b/index
.html
 Tutorial on graph cut method and
assessment of segmentation
http://www.cis.upenn.edu/~jshi/GraphT
utorial/

CSE 803 Fall 2008 Stockman 13

Segmentation via region-
growing (aggregation)

Pixels, or patches, at the lowest

level are combined when similar in
a hierarchical fashion

CSE 803 Fall 2008 Stockman 14

Decision: combine neighbors?

Neighboring
pixel or
region

CSE 803 Fall 2008 Stockman 15

Aggregation decision

CSE 803 Fall 2008 Stockman 16

Representation of regions

CSE 803 Fall 2008 Stockman 17

Chain codes for boundaries

CSE 803 Fall 2008 Stockman 18

Quad trees divide into quadrants

M=mixed; E=empty; F=full

CSE 803 Fall 2008 Stockman 19

Can segment 3D images also
 Oct trees subdivide into 8 octants
 Same coding: M, E, F used
 Software available for doing 3D image
processing and differential equations
using octree representation.
 Can achieve large compression factor.

CSE 803 Fall 2008 Stockman 20

More URLs of other work
 Mean shift description
http://cmp.felk.cvut.cz/cmp/courses/ZS
1/slidy/meanShiftSeg.pdf

CSE 803 Fall 2008 Stockman 21