Presentation is loading. Please wait.

Presentation is loading. Please wait.

CSE 803 Fall 2008 Stockman1 Ch 10 Image Segmentation Ideally, partition an image into regions corresponding to real world objects.

Similar presentations


Presentation on theme: "CSE 803 Fall 2008 Stockman1 Ch 10 Image Segmentation Ideally, partition an image into regions corresponding to real world objects."— Presentation transcript:

1 CSE 803 Fall 2008 Stockman1 Ch 10 Image Segmentation Ideally, partition an image into regions corresponding to real world objects.

2 CSE 803 Fall 2008 Stockman2 Goals of segmentation

3 CSE 803 Fall 2008 Stockman3 Segments formed by K-means

4 CSE 803 Fall 2008 Stockman4 Segmentation attempted via contour/boundary detection

5 CSE 803 Fall 2008 Stockman5 Clustering versus region-growing

6 CSE 803 Fall 2008 Stockman6 Clustering versus region-growing

7 CSE 803 Fall 2008 Stockman7 K-means clustering as before: vectors can contain color+texture

8 CSE 803 Fall 2008 Stockman8 K-means

9 CSE 803 Fall 2008 Stockman9 Histograms can show modes

10 CSE 803 Fall 2008 Stockman10 Otsu’s method assumes K=2. It searches for the threshold t that optimizes the intra class variance.

11 CSE 803 Fall 2008 Stockman11 Ohlander bifurcated the histogram recursively

12 CSE 803 Fall 2008 Stockman12 Recursive histogram-controlled segmentation

13 CSE 803 Fall 2008 Stockman13 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/Graph Tutorial/

14 CSE 803 Fall 2008 Stockman14 Segmentation via region- growing (aggregation) Pixels, or patches, at the lowest level are combined when similar in a hierarchical fashion

15 CSE 803 Fall 2008 Stockman15 Decision: combine neighbors? Neighboring pixel or region

16 CSE 803 Fall 2008 Stockman16 Aggregation decision

17 CSE 803 Fall 2008 Stockman17 Representation of regions

18 CSE 803 Fall 2008 Stockman18 Chain codes for boundaries

19 CSE 803 Fall 2008 Stockman19 Quad trees divide into quadrants M=mixed; E=empty; F=full

20 CSE 803 Fall 2008 Stockman20 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.

21 CSE 803 Fall 2008 Stockman21 More URLs of other work Mean shift description http://cmp.felk.cvut.cz/cmp/courses/ZS 1/slidy/meanShiftSeg.pdf


Download ppt "CSE 803 Fall 2008 Stockman1 Ch 10 Image Segmentation Ideally, partition an image into regions corresponding to real world objects."

Similar presentations


Ads by Google