Automated Hierarchical Image Segmentation Based on Merging of Quadrilaterals

Abstract

This paper proposes a quadrilateral-based and automated hierarchical segmentation method, in which quadrilaterals are first constructed from an edge map, where neighboring quadrilaterals with similar features of interest are then merged together in a hierarchical mode to form regions. When evaluated qualitatively and quantitatively, the proposed method outperforms three traditional and commonly-used techniques, namely, K-means clustering, seeded region growing and quadrilateral-based segmentation. It is shown by experimental results that our proposed method is robust in both recovering missed important regions while preventing unnecessary over-segmentation, and offers an efficient description of the segmented objects conducive to content-based applications.