Yi Wei

Yi Wei


Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 April 2024 in IEEE Sensors Journal
Verified
Journal Article
Closed access
YWYi Wei
Yi Wei
YXYebo Xu
Yebo Xu
WZWei Zhang
Wei Zhang
SXShaomin Xie
Shaomin Xie
YZYu Zhou
Yu Zhou
ASAo Sun
Ao Sun
In order to prompt the further application of the image measurement technique in the deformation monitoring for engineering constructions, this paper proposes a new localization solution for the accurate positioning of an artificial mark center in an image taken outdoors in a long range. Most existing localization algorithms based on image measurement only focus on the subpixel positioning part. Different from that, this paper studies the pixel level localization technique and data quality control mechanism. It firstly introduces the concept of the maximum gradient projection to extract edge pixels instead of using traditional gradient operators. Then an improved edge thinning algorithm is presented and the curvature is used to obtain an integral and reliable pixel level edge of an object with less sensitivity to the uncertain interferences. Finally, a data quality control system is designed after the subpixel localization of the mark center to select those meeting the statistical requirement. Localization errors caused by various negative effects can thus be controlled. Experimental results show that the proposed deformation monitoring technique based on the image measurement demonstrates its accurate localization performance when facing different natural environment, background and photographic ranges. It also has fast processing speed and less requirement for the imaging equipment. When using a SLR (Single Lens Reflex) with normal lens, it can achieve less than 3.5 mm localization accuracy within 200-meter photographic range. Therefore, it shows its potential in real deformation monitoring applications.

1 October 2017 in Journal of Physics: Conference Series
Verified
Conference Paper
Open access
PDF file
YWYi Wei
Yi Wei
CYChenhui Yang
Chenhui Yang
HYHang Yang
Hang Yang
YXYaming Xu
Yaming Xu
Irregular shape recognition has been a tough problem in image understanding field. One type of feature extraction techniques involves transforming shapes into signals through certain process, and thus the shape recognition problem becomes comparing similarity between signals. This paper proposes a novel signal similarity evaluation scheme that may classify signals into different categories to fulfill the shape recognition task. Experiments on pavement guide marks recognition are conducted to study the potential of the proposed method. Results demonstrate its ability in irregular shape recognition applications with higher accuracy and faster speed.

1 November 2019
Verified
Conference Paper
Closed access
SWSu Wenbo
Su Wenbo
ZYZeng Yong
Zeng Yong
DJDeng Jianan
Deng Jianan
FDFuzhou Duan
Fuzhou Duan
YWYi Wei
Yi Wei
Single imaging with plenoptic camera can simultaneously record the intensity information and angle information of the light in the scene, and generate a focal stack image through refocusing algorithm. Based on refocusing algorithm, this paper proposes an all in focus image extraction method based on fusing defocus feature and correspondence feature. Through visual qualitative evaluation and quantitative evaluations including edge intensity, information richness and sharpness, in comparison with a variety of single-featured extraction methods, the method provided in this paper shows a better performance on visual effects. There are also improvements of a certain proportion in the objective evaluation indicators.

6 February 2026
Verified
Conference Paper
Open access
TCTiantian Chen
Tiantian Chen
YWYi Wei
Yi Wei
To address the susceptibility of automated pointer gauge reading in port environments to complex backgrounds, strong illumination, and multi-scale variations, this paper proposes an improved YOLOv11n-Pose model. A High-level Screening Feature Pyramid Network (HS-FPN) is introduced to replace the original Neck structure, enhancing multi-scale feature fusion while reducing model parameters. Furthermore, a dual-path downsampling module (ADown) is designed to replace some convolutional layers in the Backbone, improving feature representation for small-scale targets and complex scenes. To enhance robustness to difficult samples such as occlusions and boundary cases, a hard sample mining mechanism (LRM) is incorporated into the loss function. Experimental results show that the proposed model achieves object detection and keypoint detection accuracies of 96.83% and 92.04%, respectively, on pointer gauge detection tasks in port environments. The model satisfies real-time and lightweight requirements and outperforms several mainstream detection methods, demonstrating its effectiveness for automated pointer gauge reading in port scenarios.

4 March 2022
Verified
Conference Paper
Closed access
LYLiu Yuansheng
Liu Yuansheng
YWYi Wei
Yi Wei
ZYZhou Yu
Zhou Yu
WJWang Jimin
Wang Jimin
SASun Ao
Sun Ao
At present, the automation of crack location in concrete wall is low, and most of them rely on manual crack location. Such manual location efficiency is low, and the error rate will be greatly improved due to long-time work. In order to solve this problem, a concrete crack location system based on transfer learning is proposed. By modifying the trained and mature deep learning model through transfer learning, a better crack location effect can be obtained in only a small number of samples and a short time. The experimental results show that the transfer learning method can effectively solve the problem of samples and time required for deep learning.

1 November 2017
Verified
Book Chapter
Closed access
PDF file
YWYi Wei
Yi Wei
LCLumeng Cao
Lumeng Cao
WYWen Yu
Wen Yu
HWHao Wu
Hao Wu