schedule

Schedule

• F5: Texture and segmentation

• F6: Energy and graph based segmentation

• F7: Active contours, snakes and level sets

• F8: Fitting, Hough transform

• F9: Recognition and classification

• …

A Vision ApplicationBinary Image Segmentation

How ?

Cost function Models our knowledge about natural images

Optimize cost function to obtain the segmentation

Why do these tokens belong together?

TRAFFIC RESEARCH

• Increase traffic safety• Increase traffic flow• Together with Traffic

Dept in Lund.• Automatic detection

and analysis of objects and events in traffic environment

K-means clustering using intensity alone and color alone

Image Clusters on intensity Clusters on color

K-means using color alone, 11 segments

Image Clusters on color

K-means usingcolor alone,11 segments.

K-means using colour andposition, 20 segments

Graph theoretic clustering

• Represent tokens using a weighted graph.– affinity matrix

• Cut up this graph to get subgraphs with strong interior links

Example eigenvector

points

matrix

eigenvector

More than two segments

• Two options– Recursively split each side to get a tree,

continuing till the eigenvalues are too small– Use the other eigenvectors