Multi-Object Segmentation by Hierarchical Layered Oriented Image Foresting Transform

Author(s)

Castaneda Leon, Leissi M. ; Miranda, Paulo A. V. ; IEEE

• Document type: Journal article
• Source: 2017 30TH SIBGRAPI CONFERENCE ON GRAPHICS, PATTERNS AND IMAGES (SIBGRAPI); v. N/A, p. 8-pg., 2017-01-01.

Abstract

This paper introduces a new method for multi-object segmentation in images, named as Hierarchical Layered Oriented Image Foresting Transform (HLOIFT). As input, we have an image, a tree of relations between image objects, with the individual high-level priors of each object coded in its nodes, and the objects' seeds. Each node of the tree defines a weighted digraph, named as layer. The layers are then integrated by the geometric interactions, such as inclusion and exclusion relations, extracted from the given tree into a unique weighted digraph, named as hierarchical layered digraph. A single energy optimization is performed in the hierarchical layered weighted digraph by Oriented Image Foresting Transform (OIFT) leading to globally optimal results satisfying all the high-level priors. We evaluate our framework in the multi-object segmentation of medical and synthetic images, obtaining results comparable to the state-of-the-art methods, but with low computational complexity. Compared to multi-object segmentation by min-cut/max-flow algorithm, our approach is less restrictive, leading to globally optimal results in more general scenarios. (AU)

• FAPESP's process: 14/50937-1 - INCT 2014: on the Internet of the Future
• Grantee: Fabio Kon
• Support Opportunities: Research Projects - Thematic Grants
• FAPESP's process: 16/21591-5 - Development of robust methods for edge delineation in images using graphs
• Grantee: Fábio Augusto Menocci Cappabianco
• Support Opportunities: Regular Research Grants
• FAPESP's process: 11/50761-2 - Models and methods of e-Science for life and agricultural sciences
• Grantee: Roberto Marcondes Cesar Junior
• Support Opportunities: Research Projects - Thematic Grants