Delimitation of regions of interest in similarity queries visualization

Author(s):Show less -

Paiva, Claudio Eduardo ; Bueno, Renato ; Banissi, E ; Ursyn, A ; Bannatyne, MWM ; Datia, N ; Francese, R ; Sarfraz, M ; Wyeld, TG ; Bouali, F ; Venturin, G ; Azzag, H ; Lebbah, M ; Trutschl, M ; Cvek, U ; Mueller, H ; Nakayama, M ; Kernbach, S ; Caruccio, L ; Risi, M ; Erra, U ; Vitiello, A ; Rossano, V

• Document type: Journal article
• Source: 2019 23RD INTERNATIONAL CONFERENCE INFORMATION VISUALISATION (IV): BIOMEDICAL VISUALIZATION AND GEOMETRIC MODELLING & IMAGING; v. N/A, p. 6-pg., 2019-01-01.

Abstract

In Content Based Image Retrieval (CBIR) systems, the visualization of queries allows to add the human visual perception in the analysis process and facilitate the discovery of knowledge. Content-based queries can be performed comparing features extracted from images, such as color, texture, and shape. In this paper we propose ways to delimit the region of interest to be visualized in the execution of queries by similarity in complex datasets. Limiting the amount of data to be visualized allows keeping the distribution of mapped data closer to the real distribution, besides allowing the application of more expensive computational methods for multidimensional projection. The proposed techniques were implemented in a prototype that allows visualizing only the region in which the query is being performed, mapping the data in three-dimensional spaces and allowing users to interact with them, being favored by human perception to improve the analysis and understanding of the data. (AU)

• FAPESP's process: 16/17078-0 - Mining, indexing and visualizing Big Data in clinical decision support systems (MIVisBD)
• Grantee: Agma Juci Machado Traina
• Support Opportunities: Research Projects - Thematic Grants