Zhengmi Tang

ORCID: 0000-0003-2011-8105
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Handwritten Text Recognition Techniques
  • Generative Adversarial Networks and Image Synthesis
  • Digital Media Forensic Detection
  • Advanced Image Processing Techniques
  • Image Processing Techniques and Applications
  • 3D Surveying and Cultural Heritage
  • Image and Object Detection Techniques
  • Image and Signal Denoising Methods
  • Advanced Vision and Imaging
  • Image Processing and 3D Reconstruction
  • Optical measurement and interference techniques
  • Industrial Vision Systems and Defect Detection
  • Structural Health Monitoring Techniques
  • Photoacoustic and Ultrasonic Imaging

Tohoku University
2021-2024

Hiroshima University
2020

Scene text erasing, which replaces regions with reasonable content in natural images, has drawn significant attention the computer vision community recent years. There are two potential subtasks scene erasing: detection and image inpainting. Both require considerable data to achieve better performance; however, lack of a large-scale real-world scene-text removal dataset does not allow existing methods realize their potential. To compensate for pairwise data, we made use synthetic after...

10.1109/tip.2021.3125260 article EN cc-by-nc-nd IEEE Transactions on Image Processing 2021-01-01

Scene-text image synthesis techniques that aim to naturally compose text instances on background scene images are very appealing for training deep neural networks due their ability provide accurate and comprehensive annotation information. Prior studies have explored generating synthetic two-dimensional three-dimensional surfaces using rules derived from real-world observations. Some of these proposed scene-text through learning; however, owing the absence a suitable dataset, unsupervised...

10.1109/tip.2023.3326685 article EN cc-by IEEE Transactions on Image Processing 2023-01-01

In this study, the dynamic deflections and vibrations of a belt conveyor system operating in ironworks are observed using high-speed telephoto mirror-drive active vision that can simultaneously switch viewpoints capture magnified images at rate hundreds frames per second (fps). The captures 160 fps pan-and-tilt scan mechanism. captured as multiple high-frame-rate video images. small vibrations, belts pillars, which have peak frequencies 10 Hz or more estimated with precision dozens...

10.2355/isijinternational.isijint-2019-643 article EN cc-by-nc-nd ISIJ International 2020-05-15

In this study, dynamic deflections and vibrations of belt conveyors operating in ironworks are observed using a high-speed telephoto mirror-drive active vision that can simultaneously switch viewpoints capture zooming-in images at hundreds frames per second. 160-fps video for conveyor captured by our system with pan-and-tilt scan as multiple high-frame-rate the experiments, small belts pillars, whose peak frequencies 10 Hz or more, estimated precision dozens micrometers image analysis such...

10.2355/tetsutohagane.tetsu-2019-059 article EN cc-by-nc-nd Tetsu-to-Hagane 2020-01-01

Background and objective: High-resolution radiographic images play a pivotal role in the early diagnosis treatment of skeletal muscle-related diseases. It is promising to enhance image quality by introducing single-image super-resolution (SISR) model into radiology field. However, conventional pipeline, which can learn mixed mapping between SR denoising from color space inter-pixel patterns, poses particular challenge for with limited pattern features. To address this issue, paper introduces...

10.48550/arxiv.2312.16455 preprint EN other-oa arXiv (Cornell University) 2023-01-01

Scene-text image synthesis techniques that aim to naturally compose text instances on background scene images are very appealing for training deep neural networks due their ability provide accurate and comprehensive annotation information. Prior studies have explored generating synthetic two-dimensional three-dimensional surfaces using rules derived from real-world observations. Some of these proposed scene-text through learning; however, owing the absence a suitable dataset, unsupervised...

10.48550/arxiv.2209.02397 preprint EN cc-by-nc-nd arXiv (Cornell University) 2022-01-01
Coming Soon ...