| Grant number: | 16/05669-4 |
| Support type: | Regular Research Grants |
| Duration: | October 01, 2016 - November 30, 2018 |
| Field of knowledge: | Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques |
| Principal Investigator: | Fabricio Aparecido Breve |
| Grantee: | Fabricio Aparecido Breve |
| Home Institution: | Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil |
Abstract
Many interactive image segmentation approaches are based on semi-supervised learning, which employs both labeled and unlabeled data in its training process. In this kind of approach, an human specialist label a few pixels from each segment, and then the semi-supervised learning algorithm labels the remaining pixels. The particle competition and cooperation model is a recent semi-supervised learning graph-based approach. It employs particles walking through a graph to classify data items which corresponde to graph nodes. Each group of particles aims to dominate the largest amount of unlabeled nodes, spreading its label and, at the same time, the particles try to avoid invasion from enemy particles in their territory. This research project main objective is to adapt the particle competition and cooperation model to perform the interactive image segmentation task. Each image pixel may be converted to a graph node and the similarity between pixels pairs, either by location or visual features, may define the edges between the corresponding nodes. Preliminary simulation results with artificial and real-world images shown that this approach is very promising. (AU)