🙋‍♀️ Profile

I obtained my Ph.D. degree from Hosei University in March 2024 and my M.S. degree in September 2020, under the supervision of Prof. Jinjia Zhou.

My research interest lies in Image Codec design based on Compressive Sensing, which encompasses Hardware Implementation and Deep Learning.

Throughout my doctoral and master’s studies, I have authored 10 papers, most of which have been published in leading conferences and journals in the fields of image processing and hardware implementation, including TMM, CVM, MMM, ISCAS, DCC, ICIP, IJCNN, among others.

I possess a strong ability for self-directed learning and have independently acquired knowledge in areas relevant to my research topic during my graduate studies, as well as in iOS application development during my undergraduate studies.

I will be joining Pengcheng Laboratory on September 9, 2024, to work on compression technology related to remote sensing.

🔥 News

• 2024.03: 🎉 One paper is accepted by IJCNN oral 2024!
• 2024.01: 🎉 One paper is accepted by CVM journal!
• 2023.10, 🎉 Keynote speech on Compressive Sensing at Institut National des Sciences Appliquées de Rennes, INSA de Rennes.
• 2023.10: 🎉 One paper is accepted by TMM!
• 2023.06: 🎉 One paper is accepted by ICIP oral 2023!
• 2023.06: 🎉 Become a reviewer for ECAI 2023!
• 2023.02: 🎉 One paper is accepted by DCC poster 2023!
• 2022.09, 🎉 Initiate recurring seminars with Fudan University and Zhejiang University.
• 2022.08, 🎉 Keynote speech on Compressive Sensing at Osaka University.
• 2022.06, 🎉 One paper is accepted by MMM oral 2022!
• 2022.05, 🎉 One paper is accepted by ISCAS oral 2022!

📝 Publications

TMM 2023

Compressive Sensing Based Image Codec With Partial Pre-Calculation

Jiayao Xu, Jian Yang, Fuma Kimishima, Ittetsu Taniguchi, Jinjia Zhou

IEEE Transaction on Multimedia, 2023 (IF = 8.182, Top Journal in Image Processing)

Key Idea:

Break the current codec framework and analyze the data characteristics within each process. Utilize the sparsity of matrices and processed data to introduce an innovative codec framework based on Compressive Sensing.

Paper / Project

MMM 2022

Real-time FPGA Design for OMP Targeting 8K image Reconstruction

Jiayao Xu, Chen Fu, Zhiqiang Zhang, Jinjia Zhou

28th International Conference on Multimedia Modeling (MMM), 2022

Key Idea:

Due to hardware limitations in sampling, we begin our work with a deterministic matrix. We utilize the sparsity of the chosen sensing matrix to streamline calculations in each iteration of reconstruction.

Experimental results demonstrate a speedup of 290 times compared to the state-of-the-art method, enabling real-time reconstruction of 8K grayscale images at 30 FPS.

Paper

ISCAS 2022

An 81.92Gpixels/s Fast Reconstruction of Images from Compressively Sensed Measurements

Jiayao Xu, Pham Do Kim Chi, Chen Fu, Jinjia Zhou

The IEEE International Symposium on Circuits and Systems (ISCAS), 2022

Key Idea:

We employ a sparse sensing matrix composed mainly of zero-vectors, resulting in an invertible full-rank matrix after removing these vectors, enabling the replacement of the iteration-based reconstruction procedure with a single matrix multiplication.

The result shows that our work accelerates the state-of-the-art method by 65 × and achieves 81.92Gpixels/s reconstruction.

Paper

Peer Review

Plug-and-Play Adaptive Block Compressive Sensing using Edge-Detection on the Compressed Domain

Jiayao Xu, Yibo Fan, Jinjia Zhou

Key Idea:

We recognize the correlation between edge detection and Compressive Sensing sampling, proposing an adaptive method that directly processes sampled data, thereby cutting time and hardware costs.

Experimental comparisons against traditional and learning-based techniques consistently reveal superior results.

ICVIP 2020

High-speed Compressed Sensing Reconstruction using Zigzag Ordering based Parallel Processing

Jiayao Xu, Jirayu Peetakul, Muchen Li, Jinjia Zhou

International Conference on Video and Image Processing (ICVIP), 2020

Key Idea:

We implement Zigzag ordering-based parallelism in the reconstruction process to ensure efficient calculations between blocks during parallel reconstruction.

Experimental results show that the proposed algorithm accelerates the baseline algorithm by 3.26 to 7.13 times.

Paper

• CVM, ACCEPT JVCSR+: Adaptively Learned Video Compressive Sensing Reconstruction with Joint In-loop Reference Enhancement and Out-loop Super-resolution Jian Yang, Jiayao Xu, Jinjia Zhou

• DCC Zigzag Ordered Walsh Matrix for Compressed Sensing Image Sensor Jinyao Zhou, Jiayao Xu, Jirayu Peetakul, Jinjia Zhou

• ICIP Dynamic Unilateral Dual Learning for Text-to-Image Synthesis Zhiqiang Zhang, Jiayao Xu, Ryugo Morita, Wenxin Yu, Jinjia Zhou

• IJCNN, ACCEPT High Frequency Feature Distillation Network for Compressive Sensing Reconstruction Fuma Kimishima, Jiayao Xu, Jinjia Zhou

• MMSP Optimizing CABAC architecture with prediction based context model prefetching Chen Fu, Heming Sun, Jiayao Xu, Zhiqiang Zhang, Jinjia Zhou

🔥 Skills

• Languages: English - TOEIC(845/990), Japanese - JLPT N2(120/180), Mandarin - Native speaker

• Programming Languages: Python, C/C++, MATLAB, Verilog , Objective-C

• Software Development: Object-Oriented Programming (OOP), Design Patterns, Team Collaboration

• Platform: Visual Studio Code, MATLAB, Modelsim, Vivado, XCode, Wireshark, Git version control

• OS (environment setting and system maintenance): Windows, Centos 7, Ubuntu 18, macOS

👩‍💻 iOS Project

Name: iSWUST (i西科)

Supported users: 3,000+ Active Users

Laboratory: Mobile Web Laboratory (移动互联网实验室)

Work duration: Sep.2015 – Mar.2017

Release Time: December 4th, 2016

Current status: out of service.

Introduction:

Provide on-campus student services such as checking lecture schedules, accessing library book collection information, verifying campus card balances, and processing recharges.

Responsibility:

1. Contributed to API design and revised API documentation to enhance comprehension and utilization by the development team.

2. Conducted a refactoring of the existing application, employing Git version control to facilitate code collaboration and project management. Utilized various design patterns to reduce coupling and enhance code maintainability.

3. Developed and integrated functionality for library and login modules, implementing rigorous testing methods to validate code accuracy and reinforce system stability.

4. Orchestrated the implementation of the network layer, encapsulating data packets in JSON format. Collaborated closely with the backend team to conduct comprehensive testing of the network layer, ensuring seamless communication with backend services and making necessary adjustments to meet project requirements.

📖 Educations

• 2020.09 - 2024.03, Ph.D, Hosei University, Tokyo, Japan.
• 2018.09 - 2020.09, Master, Hosei University, Tokyo, Japan.
• 2014.09 - 2018.06, Bachelor, Southwest University of Science and Technology, Sichuan, China.

💬 Invited Talks

• 2023.10, Compressive Sensing Based Image Codec With Partial Pre-Calculation talk in Institut National des Sciences Appliquées de Rennes, INSA de Rennes.
• 2022.09 - 2024.03, Regular seminar with Fudan University and Zhejiang University for Image Compression using Compressive Sensing.
• 2022.08, Compressive Sensing Seminar with Osaka University.
• 2022.06, Real-time hardware design for Compressive Sensing reconstruction talk, MMM 2022.
• 2022.05, a hardware design reaches 81.92 GPixel/s for Compressive Sensing reconstruction talk, ISCAS 2022.
• 2020.12, Parallel reconstruction design for Compressive Sensing reconstruction talk, ICVIP 2020.

🎖 Honors and Awards

• 2020-2023 Research grants from Hosei University Graduate School.
• 2020-2021 JASSO scholarship.
• 2020.12 The best presentation of all excellent presentations at the 3rd International Conference on Image and Video Processing (ICIVP 2020).