Improved image segmentation algorithm using graph-edges

Abstract EN

In this paper an efficient algorithm for segment digital image has been developed by measuring the evidence for a boundary between two regions in an image using (graph-edges).

The regions in the image were sorted as components, where each region in an image represents a component in the graph.

The region comparison predicate evaluates if there is evidence for a boundary between a pair of components by checking if the difference between the components, is large relative to the internal difference within at least one of the components.

A threshold function is used to control the degree the difference between components must be larger than minimum internal difference.

An important characteristic of the method is its ability to preserve detail in important image regions while ignoring detail in unimportant regions.

The classical methods depend just on external difference and ignore the internal difference, when segment two neighboring regions.