This special section of the IEEE Transactions on Visualization and Computer Graphics (TVCG) presents a curated selection of the most significant research from the 12th IEEE Pacific Visualization Symposium (PacificVis 2019), held in Bangkok, Thailand. As a premier international forum, PacificVis facilitates the exchange of innovative ideas between visualization researchers and practitioners, with a particular focus on the Asia-Pacific region.
The Selection Process
The papers featured in this section represent the pinnacle of the symposium's technical program. From a pool of 91 submissions, the guest editors employed a rigorous two-stage peer review process. Each manuscript was evaluated by a minimum of four reviewers, including members of a 46-person program committee and external experts. Ultimately, only 24 full papers were accepted for the symposium (~26% acceptance rate), and the top five (5.5%) were invited to this special journal section based on their exceptional quality and potential for impact.
Featured Research Themes
The selected papers address critical bottlenecks in modern visualization, spanning medical applications, machine learning interpretability, and large-scale data management.
Advancing Medical Precision
One highlighted study explores the transition of Augmented Reality (AR) from a novel display technology to a functional clinical tool. By investigating intuitive navigation schemes, the research demonstrates how AR can significantly improve the accuracy of needle placement during surgical procedures.
Interpreting the "Black Box" of AI
The "DeepVID" paper addresses the challenge of deep learning interpretability. By utilizing knowledge distillation, the authors provide a framework for visualizing the internal logic of image classifiers, effectively "debugging" complex neural networks to understand their decision-making processes.
Managing Information Overload
Two papers focus on the challenges of high-density data. One introduces a visual analytics interface for searching massive neural document embeddings using semantic feature selection. Another addresses visual clutter in scatter plots containing millions of points. By employing space-filling maps to "relax" dense layouts, the researchers preserve data distribution and outliers while eliminating overplotting.
Mathematical Generalization of Artistic Techniques
Bridging the gap between art and science, one paper generalizes the Linde-Buzo-Gray (LBG) stippling algorithm. This work shows how traditional artistic stippling can be mathematically repurposed to encode complex 2D scalar fields, providing a new way to represent both continuous and discrete data.
Key Insights and Future Directions
The research presented here underscores a shift toward human-centric design and observation-validating studies. The success of these methods—validated through user studies and expert case evaluations—suggests that the future of visualization lies in algorithms that directly enhance human perception and decision-making. Furthermore, the integration of the Asia-Pacific research community continues to drive innovation in handling increasingly complex and dense data types.
Summary updated