Hierarchical image analysis using irregular tessellations

Abstract

In this paper we have presented an image analysis technique in which a separate hierarchy is built over every compact object of the input. The approach is made possible by a stochastic decimation algorithm which adapts the structure of the hierarchy to the analyzed image. For labeled images the final description is unique. For gray level images the classes are defined by converging local processes and slight differences may appear. At the apex every root can recover information about the represented object in logirhtmic number of processing steps, and thus the adjacency graph can become the foundation for a reulational model of the scene.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.