Raster to Spline Conversion

Maureen Stone
Imaging Sciences Laboratory
Xerox Palo Alto Research Centers

Goal

9To come up with a minimal spline representation for a set of sampled points on an edge.

Approach

9Scanned high contrast images are contoured to get edge information. (100 points/inch, 256 grey levels)

9Contour is analyzed with respect to curvature.

9Information is used to place cusps and knots for spline fitting.