Ling Yu

Dr. Ling Yu

orcid

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

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

Representative Publications

View All Publications
10 June 2016 in Circuits, Systems, and Signal Processing
Verified
Journal Article
Closed access
PDF file
LYLing Yu
Ling Yu
TQTianshuang Qiu
Tianshuang Qiu
ASAi-Min Song
Ai-Min Song

1 May 2025 in AIP Advances
Verified
Journal Article
Open access
TZTianyi Zheng
Tianyi Zheng
LYLing Yu
Ling Yu
YSYongbao Shi
Yongbao Shi
FNFanglin Niu
Fanglin Niu
In the steel production process, surface defect detection is crucial for ensuring product quality. To address the issues of high computational cost and low detection accuracy in current steel defect detection models, we propose a YOLOv9-based steel defect detection algorithm, CCSS-YOLO. First, we integrate the SENetV2 attention mechanism, which has a multi-branch structure and compression and excitation operations, into the YOLOv9 feature module RepNCSPELAN4 to form a new RepNCSPELAN4_SENetV2 feature extraction module, enhancing the model’s ability to extract complex features. Next, we replace the regular convolution blocks in the model network with spatial-to-depth convolutions, further reducing the model’s computational complexity while retaining global feature information. Finally, we replace the Fusion module in the CNN-based cross-scale feature fusion (CCFM) module, with the new Fusion-RepNCSPELAN4 module, creating a new feature fusion network, CCFM-YOLO, which replaces the neck network of YOLOv9. This approach improves the model’s feature extraction capability while reducing its parameter count. Experimental results show that, on the NEU-DET dataset, CCSS-YOLO improves average detection accuracy by 1.9% compared to the original YOLOv9 model, reduces parameters by 25.4%, and decreases FLOPs by 15.6%. In addition, it demonstrates good generalization performance on the new ASVD washer dataset.

30 December 2024 in Neurocomputing
Verified
Journal Article
Open access
ZJZijian Jiang
Zijian Jiang
LYLing Yu
Ling Yu
YHYu Han
Yu Han
JLJunru Li
Junru Li
FNFanglin Niu
Fanglin Niu

13 January 2022 in Journal of Sensors
Verified
Journal Article
Open access
PDF file
HZHang Zhang
Hang Zhang
FNFanglin Niu
Fanglin Niu
LYLing Yu
Ling Yu
SZSi Zhang
Si Zhang
In traditional wireless sensor networks, information transmission usually uses data encryption methods to prevent information from being stolen illegally. However, once the encryption methods are leaked, eavesdropping nodes can easily obtain information. LT codes are rateless codes; if it is attacked by random channel noise, the decoding process will change and the decoding overhead will also randomly change. When it is used for physical layer communication of wireless sensor networks, it ensures that the destination node recovers all the information without adding the key, while the eavesdropping node can only obtain part of the information to achieve wireless information security transmission. To reduce the intercept efficiency of eavesdropping nodes, a physical layer security (PLS) method of LT codes with double encoding matrix reorder (DEMR-LT codes) is proposed. This method performs two consecutive LT code concatenated encoding on the source symbol, and part of the encoding matrix is reordered according to the degree value of each column from large to small, which reduces the probability of eavesdropping nodes recovering the source information. Experimental results show that compared with other LT code PLS schemes, DEMR-LT codes only increase the decoding overhead by a small amount. However, it can effectively reduce the intercept efficiency of eavesdropping nodes and improve information transmission security.

30 May 2022 in IEEE Access
Verified
Journal Article
Open access
SZSi Zhang
Si Zhang
FNFanglin Niu
Fanglin Niu
LYLing Yu
Ling Yu
YZYongxiang Zhang
Yongxiang Zhang
In recent years, physical layer security technology has provided a new way to solve the problem of secure transmissions in wireless communications from an information theory perspective by using physical layer security coding and the inherent characteristics of the channel. As a nonsystematic code, Luby transform (LT) code has the characteristics of random coding and rateless, therefore, eavesdroppers cannot directly obtain useful information from leaked code symbols. As a shifted LT (SLT) code, the SLT codes can efficiently recover information, with a smaller decoding overhead. Therefore, we propose an SLT-LT joint code anti-eavesdropping scheme based on random symbol sets. Based on the environment of the main channel, the sender selects a set of random symbols to send to legitimate receivers as known symbols and uses these random symbols and message symbols to form source symbols for the SLT-LT concatenated coding to increase the bit error rate (BER) of the eavesdroppers. The experimental results show that, compared with other anti-eavesdropping LT schemes, our proposed scheme only adds a small amount of decoding overhead but has a better security performance.

12 June 2025 in PLOS ONE
Verified
Journal Article
Open access
PDF file
RLRunqing Li
Runqing Li
LYLing Yu
Ling Yu
ZJZijian Jiang
Zijian Jiang
FNFanglin Niu
Fanglin Niu
The RGB-D salient object detection technique has garnered significant attention in recent years due to its excellent performance. It outperforms salient object detection methods that rely solely on RGB images by leveraging the geometric morphology and spatial layout information from depth images. However, the existing RGB-D detection model still encounters difficulties in accurately recognising and highlighting salient objects when facing complex scenes containing multiple or small objects. In this study, a Cross-modal Interactive and Global Awareness Fusion Network for RGB-D Salient Object Detection, named CIGNet, is proposed. Specifically, convolutional neural networks (CNNs), which are good at extracting local details, and an attention mechanism, which efficiently integrates global information, are utilized to design two fusion methods for RGB and depth images. One of these methods, the Cross-modal Interaction Fusion Module (CIFM), employs depth separable convolution and common-dimensional dynamic convolution to extract rich edge contours and texture details from low-level features. The Global Awareness Fusion Module (GAFM) is designed to relate high-level features between RGB and depth features so as to improve the model’s understanding of complex scenes. In addition, prediction mapping is generated through a step-by-step decoding process carried out by the Multi-layer Convolutional Fusion Module (MCFM), which gradually yields finer detection results. Finally, comparing 12 mainstream methods on six public benchmark datasets demonstrates superior robustness and accuracy.