Data Science
Hierarchical Image Segmentation - Research

Overview

Image segmentation is the partitioning of an image into related sections or regions. For instance, a remotely sensed image of an area with distinct surface features could be analyzed and sorted by regions that contain snow, water, human infrastructure, etc. Segmentation is a key first step for a number of different approaches to image analysis.

A segmentation hierarchy is useful for applications that require different levels of image segmentation detail depending on the particular image objects segmented. A unique feature of a segmentation hierarchy is that the segment or region boundaries are maintained at the full image resolution for all levels of the segmentation hierarchy. An object of interest may be represented by multiple image segments in finer levels of detail, or may be included in a surrounding region at coarser levels of detail in the segmentation hierarchy. If the segmentation hierarchy has sufficient numbers of hierarchical levels, the object will be delineated as one region or region component at an intermediate level in the segmentation hierarchy.

Approaches for producing an image segmentation hierarchy usually fall into two categories: (1) multiresolution analysis and (2) region growing. A segmentation hierarchy can be generated directly by analyzing the multiresolution images. A drawback to using multiresolution methods to obtain a segmentation hierarchy is the complexity and, perhaps, the computational requirements of these approaches.

The Hierarchical Image Segmentation (HSEG) approach described here is of the region growing category. The development of HSEG and its recursive formulation (RHSEG) is described in the following sections:

  • Approaches to Region Growing Segmentation
  • Improvements in Hierarchical Step-Wise Optimization
  • Hierarchical Image Segmentation
  • HSEG Software