Research

Research Keywords

  • Information Engineering, Computer Science, Computational Imaging
  • Multi-view 3D Image/Video, Cross Reality (XR), 3D Displays
  • 3D CG, Shitsukan (Material Perception) Informatics, StreamDiffusion, Avatar Quality Enhancement
  • Image Media Quality, Visual Information Science, Neural Networks
  • Image/Signal Processing, Digital Watermarking, Information Security (AES Encryption, Scrambling)
  • Color Information Engineering, Image Analysis, Color Management (S-CIELAB Color Space)
  • Image Coding (H.265/HEVC), Super-Resolution (SRCNN/GAN), Total Variation Regularization
  • Statistical Analysis, Data Science, Machine Learning, Vision Transformer, Zero-Shot Learning
  • Human-Interface Interaction, Medical VR, Autonomous Driving (Autoware)
  • Medical Imaging, Computer-Aided Diagnosis (CAD), Museum DX, Digital Archives

Research Fields

  • Informatics / Perceptual information processing / Image Processing & Visual Media Processing & Media Database
  • Informatics / Intelligent informatics / Machine Learning & Intelligent Information Processing & Data Mining
  • Informatics / Biological, health, and medical informatics / Biological Information & Health Information & Medical Image
  • Life sciences / Medical systems / Diagnostic Imaging System & Medical Information System & Computer Aided Surgery
  • Manufacturing technology (mechanical, electrical/electronic, chemical engineering) / Communication and network engineering & Signal Processing & Network & Multimedia Communication
  • Informatics / Human interfaces and interactions / Virtual Reality & Usability & Ergonomics & Human Factor
  • Informatics / Information theory / Information Theory & Coding Theory & Learning Theory
  • Informatics / Statistical science / Data Science & Modeling & Multivariate Analysis
  • Informatics / Computer systems / Circuits and System & Hardware and Software co-design & Embedded System
  • Informatics / Entertainment and game informatics / 3D Contents & Animation & Media Art

Affiliated Societies

  • The Institute of Image Information and Television Engineers (ITE) (Full Member, Feb. 2011 – Present)
  • The Institute of Electronics, Information and Communication Engineers (IEICE) (Full Member, Aug. 2013 – Present)
  • The Institute of Electrical and Electronics Engineers, Signal Processing Society (IEEE SPS) (Member, Feb. 2014 – Present)
  • The Institute of Image Electronics Engineers of Japan (IIEEJ) (Full Member, Apr. 2015 – Present)
  • Information Processing Society of Japan (IPSJ) (Full Member, Feb. 2017 – Present)
  • The Japanese Society of Medical Imaging Technology (JAMIT) (Full Member, Apr. 2018 – Present)
  • Japan Society of Computer Aided Surgery (JSCAS) (Full Member, Jul. 2018 – Present)
  • The International Society for Optical Engineering (SPIE) (Early Career Professional Member, Jun. 2020 – Present)
  • IEEE Engineering in Medicine and Biology Society (IEEE EMBS) (Member, Oct. 2020 – Present)
  • IEEE Computer Society (IEEE CS) (Member, Oct. 2020 – Present)
  • Japan Association for Medical Informatics (JAMI) (Full Member, Aug. 2021 – Present)
  • Society for Imaging Science and Technology (IS&T) (Member, Dec. 2021 – Present)
  • Japan DataScientist Association (Member, Jun. 2025 – Present)
  • The Institute of Electrical Engineers of Japan (IEEJ) (Member, Sep. 2025 – Present)
  • The Color Science Association of Japan (CSAJ) (Full Member, Feb. 2026 – Present)

* Student membership for all except IEEE SPS until March 2017; Student Member for IEEE SPS until March 2017.

Research Overview

1. Ergonomic Studies on Quality Assessment for Multi-view 3D Displays and High-Definition Images

With the advancement of super-reality systems and digital signage, 3D CG-based stereoscopic video content has become widespread. This research evaluates the impact of camera intervals and view-dependent coding degradation on viewers’ perception of stereoscopic depth, sense of enclosure, and material texture across various 3D formats, including active shutter glasses, 8-viewpoint lenticular lenses, and parallax barrier systems. Furthermore, ergonomic statistical analyses factoring in visual performance and individual characteristics (e.g., age, gender, and vision correction) were conducted, establishing a foundation for objective quality metrics aimed at constructing optimized viewing environments.

2. Integration of Image Coding and Visible Digital Watermarking in Multi-view 3D CG Images

Digital watermarking is critical for intellectual property protection; however, applying it to multi-view 3D images requires balancing visual quality with disparity and stereoscopic depth perception. This research investigates the optimal application sequence of H.265/HEVC coding and visible digital watermarking (leveraging wavelet transforms), as well as perceptual tolerance regarding different RGB components and frequency domains. By utilizing machine learning models such as Support Vector Machines (SVM), we achieve advanced classification of image quality and embedding strength patterns, quantitatively and qualitatively clarifying the prerequisites for secure, high-quality media communications.

3. Medical AI Information Systems and Automatic Computer-Aided Diagnosis for Color Laparoscopic Videos

Aiming to prevent diagnostic oversights and enhance accuracy in medical fields, we implement artificial intelligence techniques into Computer-Aided Diagnosis (CAD) systems. Focusing on intraoperative color laparoscopic videos, this study evaluates the optimal design of super-resolution processing via Generative Adversarial Networks (GAN) and analyzes how contrast enhancement methodologies affect semantic segmentation precision using wavelet multi-resolution analysis. This work facilitates the realization of real-time, large-scale automated surgical decision support systems.

4. Optimal Image Representation and Noise Removal via Sparse Dictionary Learning and Total Variation Regularization

To reconcile efficient sparse representation with advanced noise reduction, we develop algorithms utilizing mathematical optimization approaches. For degradation stemming from image coding and transmission noise, we apply sparse dictionary coding and Total Variation (TV) regularization to effectively suppress noise while preserving fine texture details. Through publications in journals like Displays and technical articles in Image Lab (Nippon Industrial Publishing), we expand these applications to broadcasting and medical imaging workflows.

5. Semantic Segmentation and Zero-Shot Learning Applications for 3D CG Images

We quantitatively evaluate how high-compression coding (e.g., H.265/HEVC) affects automated scene understanding and image recognition powered by AI. Utilizing 3D CG image datasets, this research analyzes changes in semantic segmentation precision under the presence of compression artifacts. Recently, we have integrated “Zero-Shot Learning" to enable flexible object segmentation of unseen classes, driving forward advanced environmental perception through the fusion of image engineering and state-of-the-art deep learning paradigms.

6. Visual Attention and Region of Interest (ROI) Estimation Using Saliency Maps

This research models the mathematical mechanisms of human “Visual Attention" for engineering applications. By calculating “Saliency Maps" that predict where the human eye focuses first within 3D CG or laparoscopic medical images, we automatically estimate the Region of Interest (ROI). This methodology is applied to ROI-driven image compression—maintaining high quality only in perceptually critical regions—and improves the guidance accuracy of surgical support systems based on rigorous statistical analysis.

7. Self-Position Estimation and Point Cloud Analysis for Autonomous Vehicles Using Autoware

To enhance safety infrastructures in next-generation transportation systems, we conduct extensive simulation research using “Autoware," an open-source autonomous driving software. We approach high-precision self-position estimation (localization)—a critical factor in accident prevention—from the dual perspectives of information and image engineering. By performing detailed analyses on 3D point cloud data (point cloud) density and associated noise patterns, we advance the development of holistic usability architectures and hardware-software co-design frameworks.

8. Quantum Information Computing and Generative AI Quality Metrics for 360-Degree Cameras and XR

We pioneer next-generation video media processing frameworks tailored for Cross Reality (XR) spaces and the metaverse. This involves leading foundational research on introducing “Quantum Information Computing" into the data compression pipelines of 360-degree cameras. Additionally, we construct frameworks for both objective and subjective quality assessment of 3D CG images synthesized by text-to-image generative AI models like Stable Diffusion. Our recent work also evaluates real-time quality metrics and ultra-low latency enhancement for 3D CG avatars using StreamDiffusion.

9. Digital Archives and Secure Utilization Technologies of Historical Materials for Museum DX

As a core research pillar at the Kanazawa University Emerging Media Initiative, we develop advanced digital archiving methods and secure utilization mechanisms to preserve cultural assets and historical materials for future generations. Utilizing high-fidelity image analysis to capture precise material textures, we simultaneously design secure information infrastructures—incorporating AES encryption and advanced scrambling algorithms—to prevent unauthorized copying or falsification, driving forward the social implementation of Museum Digital Transformation (DX).