Tao Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tao Liu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
Zechen Li
Chengdu University of Information Technology
2 shared publicationsCan Lai
Chengdu University of Information Technology
1 shared publication- Dongdong Yang
Soochow University
1 shared publication - Haijiang Wang
College of Electronic Engineering, Chengdu University of Information Technology, Chengdu 610225, China
1 shared publication - Jiamin Wang
College of Electronic Engineering, Chengdu University of Information Technology, Chengdu 610225, China
1 shared publication - Jieyi Li
Guangzhou Meteorological Integrated Security Center, Guangzhou 511430, China
1 shared publication
Latest Publications
- Open Access
- 29 May 2026
- Published by MDPI AG in Remote Sensing
- Vol. 18 (11) , 1743
Joint Hail Detection from Satellite and Radar Observations with Spatially Adaptive Alignment and Wavelet-Gated Refinement
Detecting hail from remote sensing observations remains challenging because hail develops rapidly and its signatures may appear at different levels within a storm. Ground-based radar and geostationary meteorological satellites are the two primary observing systems for this task, yet their observations are often spatially misaligned. Satellite measurements mainly characterize the thermal structure near the cloud top, whereas radar observations capture the lower-level precipitation core. This mismatch is further exacerbated by satellite parallax, namely the apparent horizontal shift of high cloud tops caused by the oblique viewing geometry of a geostationary satellite, together with the vertical tilt of convective storms. Existing joint methods generally combine satellite cloud-top information with radar precipitation information directly, without explicitly correcting the spatial displacement, which limits detection accuracy. To address this issue, we propose HailDeformer, a deep learning framework that first aligns satellite and radar features through a bidirectional deformable cross-attention module equipped with a position-wise confidence gate and optimized with smoothness, contrastive alignment, and observation-structure consistency losses, and then refines the fused representation using an inter-scale attention module and a wavelet-guided refinement module. Experiments on a four-region dataset from China show that HailDeformer consistently outperforms Direct Fusion, Manual Weighting, Cross-Attention Fusion, and Optical Flow Alignment, achieving a mean Average Precision at IoU 0.5 (mAP@0.5) of 0.916, an F1 score of 0.864, a Critical Success Index (CSI) of 0.760, and the lowest False Alarm Ratio (FAR) of 0.149. Ablation studies further confirm that all proposed modules and associated constraints contribute to the overall performance, with the alignment module providing the largest improvement. Additional evaluations demonstrate that HailDeformer remains effective throughout storm evolution and under challenging observational conditions.
- 29 May 2026
- Published by MDPI AG in Remote Sensing
- Vol. 18 (11) , 1743
Open AccessJoint Hail Detection from Satellite and Radar Observations with Spatially Adaptive Alignment and Wavelet-Gated Refinement
Detecting hail from remote sensing observations remains challenging because hail develops rapidly and its signatures may appear at different levels within a storm. Ground-based radar and geostationary meteorological satellites are the two primary observing systems for this task, yet their observations are often spatially misaligned. Satellite measurements mainly characterize the thermal structure near the cloud top, whereas radar observations capture the lower-level precipitation core. This mismatch is further exacerbated by satellite parallax, namely the apparent horizontal shift of high cloud tops caused by the oblique viewing geometry of a geostationary satellite, together with the vertical tilt of convective storms. Existing joint methods generally combine satellite cloud-top information with radar precipitation information directly, without explicitly correcting the spatial displacement, which limits detection accuracy. To address this issue, we propose HailDeformer, a deep learning framework that first aligns satellite and radar features through a bidirectional deformable cross-attention module equipped with a position-wise confidence gate and optimized with smoothness, contrastive alignment, and observation-structure consistency losses, and then refines the fused representation using an inter-scale attention module and a wavelet-guided refinement module. Experiments on a four-region dataset from China show that HailDeformer consistently outperforms Direct Fusion, Manual Weighting, Cross-Attention Fusion, and Optical Flow Alignment, achieving a mean Average Precision at IoU 0.5 (mAP@0.5) of 0.916, an F1 score of 0.864, a Critical Success Index (CSI) of 0.760, and the lowest False Alarm Ratio (FAR) of 0.149. Ablation studies further confirm that all proposed modules and associated constraints contribute to the overall performance, with the alignment module providing the largest improvement. Additional evaluations demonstrate that HailDeformer remains effective throughout storm evolution and under challenging observational conditions.
- 23 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. PP (15) , 1
Triple-Channel Time-Frequency Fusion Custom Neural Network with OpenMax for Open-Set Recognition of Communication Signal Modulations
With the continuous advancement of wireless communication technologies and the growing diversity of modulation schemes, Automatic Modulation Recognition (AMR) plays a crucial role in spectrum sensing and signal analysis. However, traditional deep learning models with fixed architectures often face challenges in simultaneously achieving high-precision classification of known modulation types and effectively rejecting unknown modulated signals. To address this, we propose a novel framework combining a custom neural network with OpenMax. First, we introduce a time-frequency diagram channel alongside the I/Q channels to capture both time-domain waveform details and spectral structure, which enhances the model's ability to distinguish between complex modulation schemes and detect unknown signals in low-SNR environments. The network input consists of the time-frequency diagram, I component, and Q component. Moreover, the OpenMax open-set recognition model is incorporated to perform tail probability fitting and re-calibration of the neural network's activation vectors, enabling precise classification of known categories and dynamic rejection of unknown signals. Extensive simulations under challenging conditions—including Rayleigh fading, carrier frequency offset, and AWGN with SNR ranging from ‑10 dB to 20 dB—demonstrate that the proposed method achieves a closed-set classification accuracy of 92.21% and an unknown signal detection rate of 84.38%. Furthermore, computational complexity and runtime analysis show that the proposed framework achieves end-to-end inference latencies of approximately 2.1 ms on GPU and 3.7 ms on CPU, including time-frequency preprocessing, confirming its practical feasibility for real-time deployment. Compared to conventional methods like dual-channel CNN, dual-channel/triple-channel ResNet, and CNN+LSTM, the proposed approach demonstrates quite good performance in both closed-set and open-set tasks, effectively validating the effectiveness of the time-frequency fusion strategy and uncertainty calibration mechanism in improving model robustness and generalization.
- 23 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. PP (15) , 1
Triple-Channel Time-Frequency Fusion Custom Neural Network with OpenMax for Open-Set Recognition of Communication Signal Modulations
With the continuous advancement of wireless communication technologies and the growing diversity of modulation schemes, Automatic Modulation Recognition (AMR) plays a crucial role in spectrum sensing and signal analysis. However, traditional deep learning models with fixed architectures often face challenges in simultaneously achieving high-precision classification of known modulation types and effectively rejecting unknown modulated signals. To address this, we propose a novel framework combining a custom neural network with OpenMax. First, we introduce a time-frequency diagram channel alongside the I/Q channels to capture both time-domain waveform details and spectral structure, which enhances the model's ability to distinguish between complex modulation schemes and detect unknown signals in low-SNR environments. The network input consists of the time-frequency diagram, I component, and Q component. Moreover, the OpenMax open-set recognition model is incorporated to perform tail probability fitting and re-calibration of the neural network's activation vectors, enabling precise classification of known categories and dynamic rejection of unknown signals. Extensive simulations under challenging conditions—including Rayleigh fading, carrier frequency offset, and AWGN with SNR ranging from ‑10 dB to 20 dB—demonstrate that the proposed method achieves a closed-set classification accuracy of 92.21% and an unknown signal detection rate of 84.38%. Furthermore, computational complexity and runtime analysis show that the proposed framework achieves end-to-end inference latencies of approximately 2.1 ms on GPU and 3.7 ms on CPU, including time-frequency preprocessing, confirming its practical feasibility for real-time deployment. Compared to conventional methods like dual-channel CNN, dual-channel/triple-channel ResNet, and CNN+LSTM, the proposed approach demonstrates quite good performance in both closed-set and open-set tasks, effectively validating the effectiveness of the time-frequency fusion strategy and uncertainty calibration mechanism in improving model robustness and generalization.
- Open Access
- 4 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 150186-150198
ADS-B Based Wind Speed Vector Inversion Algorithm
Besides collecting and broadcasting aviation surveillance parameters, Automatic Dependent Surveillance-Broadcast (ADS-B) is also a novel technique to sense and share meteorological information such as wind field with high update rate and accuracy. Many ADS-B devices on aircraft can construct a real-time and dynamic sensor network. Although the ADS-B message format reserves the items especially for wind information, few aircrafts broadcast these items at present. The current solution is to use the aircraft trajectory captured by ADS-B for wind vector inversion. Nevertheless, some algorithms still have some downsides, especially the stability in small-angle turning situations. This paper is committed to developing a novel algorithm capable of working in both small and large angle turning situations with high efficiency, with an emphasis on small angle situations. By virtue of the algorithm in our recent research which is derived from the Particle Filter model, this algorithm takes advantage of circle geometry property and Euclidean distance standard deviation (STD). In the simulation test, the effect of true airspeed (TAS) difference on the mean absolute error of wind estimate, the effect of true wind speed on wind estimate, the effect of maneuver turning angle on wind estimate, and the computational complexity are assessed, respectively. Moreover, for real ADS-B data, both large and small-angle turning maneuver situations are tested and compared separately. Also compared is the level of the results concentration for the wind speed and TAS along with the geometric height. Consequently, the simulation and the real data test shows that the proposed STD algorithm has performance superior to other two algorithms LS and LM especially in the small-angle turning situation such as below 40deg. STD’s performance is between other two algorithms in computational complexity. This property can help improve the algorithm’s stability and data utilization for small-angle turning significantly, which is very useful in real aviation surveillance operations.
- 4 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 150186-150198
Open AccessADS-B Based Wind Speed Vector Inversion Algorithm
Besides collecting and broadcasting aviation surveillance parameters, Automatic Dependent Surveillance-Broadcast (ADS-B) is also a novel technique to sense and share meteorological information such as wind field with high update rate and accuracy. Many ADS-B devices on aircraft can construct a real-time and dynamic sensor network. Although the ADS-B message format reserves the items especially for wind information, few aircrafts broadcast these items at present. The current solution is to use the aircraft trajectory captured by ADS-B for wind vector inversion. Nevertheless, some algorithms still have some downsides, especially the stability in small-angle turning situations. This paper is committed to developing a novel algorithm capable of working in both small and large angle turning situations with high efficiency, with an emphasis on small angle situations. By virtue of the algorithm in our recent research which is derived from the Particle Filter model, this algorithm takes advantage of circle geometry property and Euclidean distance standard deviation (STD). In the simulation test, the effect of true airspeed (TAS) difference on the mean absolute error of wind estimate, the effect of true wind speed on wind estimate, the effect of maneuver turning angle on wind estimate, and the computational complexity are assessed, respectively. Moreover, for real ADS-B data, both large and small-angle turning maneuver situations are tested and compared separately. Also compared is the level of the results concentration for the wind speed and TAS along with the geometric height. Consequently, the simulation and the real data test shows that the proposed STD algorithm has performance superior to other two algorithms LS and LM especially in the small-angle turning situation such as below 40deg. STD’s performance is between other two algorithms in computational complexity. This property can help improve the algorithm’s stability and data utilization for small-angle turning significantly, which is very useful in real aviation surveillance operations.
- 4 April 2020
- Published by Springer Nature
- p. 448-458
A Triangular Centroid Location Method Based on Kalman Filter
- 4 April 2020
- Published by Springer Nature
- p. 448-458
A Triangular Centroid Location Method Based on Kalman Filter
- Open Access
- 6 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 9196-9208
The Application of the Machine Learning Method in Electromyographic Data
- 6 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 9196-9208
Open AccessThe Application of the Machine Learning Method in Electromyographic Data