Analyzing the Effects of Image Segmentation in Compound Structures

Abstract

Image segmentation is a vital part in today’s technological changes. We proposed an algorithm for deriving the compound structure
which consists of primitive image objects. Our algorithm works in two phases. Firstly it produces image regions with homogeneous spectral
content using the segmentation process. Secondly the segmentation is translated into relational graph structure .In relational graph structure each
node represents the specified region and the edges between them acts as a relation between the regions. The region objects that appear together
frequently can be considered as strongly related. Experiments are carried out using an Ikonos image which shows that subgraphs found within
the graph representing the whole image correspond to parts of different high-level compound structures.

Key words: Compound structure, Image segmentation, object detection, graph-based analysis.