研究発表

In this talk, the latest results of the Computational Forensics project were introduced with focus on the systems FORSETI and THEMIS.

Prof. Tosiyasu L. Kunii passed away in November 2020. He had been a professor at Department of Information Sciences, Faculty of Science, the University of Tokyo and was the founding president of the University of Aizu. CGI 2021 held a workshop in memory of Prof. Kunii since the conference was founded by him in 1987. Fujishiro started his carrier as an assistant professor to work with Prof. Kunii at the University of Tokyo in 1985 and hence, he was invited to provide a talk for the workshop. Fujishiro picked up a topic of FORSETI, a computational forensic visualization system under development. It is because in 1985, the first paper coauthored by Prof. Kunii and Fujishiro proposed a medical imaging method. The paper was published in the second volume of the Visual Computer journal, which was also initiated by Prof. Kunii as the first editor in chief in 1985. Following Fujishiro’s brief message in homage to Prof. Kunii, the main body of the talk was delivered by Baoqing Wang (D3). The original paper on FORSETI is to be published in a forthcoming issue in memory of Prof. Kunii of the Visual Computer Journal.

This is a conference presentation of the paper with the same title, published in IEEE Transactions on Visualization and Computer Graphics.

研究室で2021年度に発表したCG/可視化技術の事例6例をまとめて紹介した．

総合知を獲得するうえで可視化が果たす役割に関して，登壇発表者を中心にパネル討論を行った．講演ビデオ：https://www.arc.ritsumei.ac.jp/j/news/pc/012773.html

第7回の日本学術会議可視化公開シンポジウム．可視化の新パラダイム策定小委員会委員長として，招待講演を行った．講演ビデオ：https://www.arc.ritsumei.ac.jp/j/news/pc/012773.html

Many problems are observed in the form of time-varying multi-dimensional datasets. In this lecture, we will discuss how large datasets of this kind are visually explored, with a particular focus on two kinds of dynamic visual querying, namely, query by example and query by sketch. These querying approaches serve as more efficient ways of identifying observable and recurring time variation patterns in long-term datasets. Dynamic visual querying is amplified with an auxiliary mechanism called “fact-guided querying,” which is intended to iteratively refine queries and results by specifying a query on the basis of the findings of a previous query. The efficacy of these concepts will be illustrated with actual analysis examples executed in a visual analytics environment called TimeTubeX for blazar observation research.

Inspired by the trick artworks of Hideyuki Nagai, a young Japanese artist, we have developed a simple naked-eye stereoscopic viewing system for personal use. It induces motion parallax through the orthogonal arrangement of two general-purpose display monitors and combining anamorphosis, known as monocular illusion expression, with tracking the viewer’s eyes. However, the stereoscopic effect perceived by the viewer can be degraded when using small display monitors due to binocular disparity. In this talk, I sketch our latest idea of pinpointing the anamorphosis effect by identifying the different degrees of contribution the two eyes make when recognizing the composition of a synthetic space. We have empirically proved that rather than detecting and tracking the dominant eye, the sense of presence is possibly improved by placing a so-called Cyclopean eye at a certain internal position between the two eyes to obtain the pinpointed anamorphosis. We report on a trial porting of this proposed approach to a commercially available laptop PC with a foldable display and present a few examples of immersive cyberworlds.

The following three topics were discussed: 1. To extend the influence of the Asian visualization community in the visualization field, besides of ChinaVis and the Journal of Visualization, how can we further explore the collaboration possibilities between CSIG-VIS and the visualization community in Japan? 2. Can you discuss the interaction between visualization and other domain fields. For instance, what implications does the development of AI have in the field of visualization? How can visualization and visual analytics be applied in emerging topics like Tokyo Olympics, Peking winter Olympics, and the Covid-19 pandemic? 3. Do you have any suggestions for visualization researchers, especially those doing scientific visualization on how shall we adopt display and interaction modalities. For instance, how to develop visualizations in VR/AR environment? How shall we conduct user studies to make the visualizations more convincing?

This panel focuses mainly on possible collaborative research between China and Japan.

Many problems are observed in the form of time-varying multi-dimensional datasets. In this lecture, we will discuss how to visually explore large datasets of that kind, with a particular focus on automatic feature extraction, dynamic visual querying, and clustering. Automatic feature extraction is dedicated to the discovery of dynamic time variations over multiple variables and their characteristic derivatives. Dynamic visual querying provides more efficient exploration of recurring time variation patterns and subsequences to validate specific hypotheses. Clustering helps to identify hidden universalities in a long-term dataset by classifying subsequences according to their similarities. These concepts will be illustrated with several examples executed in a system, called TimeTubeX, for blazar observation research.

研究室で2019～2020年度に発表した可視化技術の事例をまとめて紹介した．