Correction of Width of Regions Using Compactness

Abstract

A new method for calculating widths of regions in a segmented image is proposed. Width is one of invariant features of a region, and can be calculated using the boundary length and the area. However, the value may be affected by the shape of the region. We introduce a function to correct the error according to the shape using compactness, which is a feature showing the shape of the region. We made models for region shapes with a rectangle and a capsule-shape, and evaluated the error to be cancelled. Results for actual data are shown.