Fast Hierarchical Image Segmentation through Irregular Pyramids

G. Valencia, J.A. Rodríguez, C. Urdiales, A. Bandera, and F. Sandoval (Spain)

Keywords

hierarchical segmentation, multilevel quadtree structure, irregular pyramid

Abstract

This paper presents a new hierarchical segmentation algorithm based on a pyramidal structure. Nodes at the different levels of this pyramid are related to connected regions at the base presenting an homogeneous gray-level. The pyramid is built in a bottom-up way and then stabilized in a top-down non-iterative way to split the image into homogeneous regions. The main advantages of the proposed algorithm is that it is computationally very fast and also that it preserves region connectivity. Besides, the complexity of the resulting decomposition only depends on the complexity of the image and not on the relative positions of the different objects in the image.

Important Links:



Go Back


IASTED
Rotating Call For Paper Image