Linlin Xu

Linlin Xu

orcid

Research Focus & Metrics

Academic Career Length

21Years (2006-Present)

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

Representative Publications

View All Publications
18 August 2022 in IEEE Transactions on Intelligent Transportation Systems
Verified
Journal Article
Closed access
DLdening lu
dening lu
QXQian Xie
Qian Xie
Kyle GaoKyle Gao
Kyle Gao
Linlin XuLinlin Xu
Linlin Xu
Jonathan Li Jonathan Li
Jonathan Li
Point cloud classification is a fundamental task in 3D applications. However, it is challenging to achieve effective feature learning due to the irregularity and unordered nature of point clouds. Lately, 3D Transformers have been adopted to improve point cloud processing. Nevertheless, massive Transformer layers tend to incur huge computational and memory costs. This paper presented a novel hierarchical framework that incorporated convolutions with Transformers for point cloud classification, named 3D Convolution-Transformer Network (3DCTN). It combined the strong local feature learning ability of convolutions with the remarkable global context modeling capability of Transformers. Our method had two main modules operating on the downsampling point sets. Each module consisted of a multi-scale local feature aggregating (LFA) block and a global feature learning (GFL) block, which were implemented by using the Graph Convolution and Transformer respectively. We also conducted a detailed investigation on a series of self-attention variants to explore better performance for our network. Various experiments on ModelNet40 and ScanObjectNN datasets demonstrated that our method achieves state-of-the-art classification performance with a lightweight design. The code is publicly available at https://github.com/d62lu/3DCTN.

5 July 2025 in ISPRS Journal of Photogrammetry and Remote Sensing
Verified
Journal Article
Open access
ZXZhengsen Xu
Zhengsen Xu
JLJonathan Li
Jonathan Li
SCSibo Cheng
Sibo Cheng
XRXue Rui
Xue Rui
YZYu Zhao
Yu Zhao
Linlin XuLinlin Xu
Linlin Xu

16 October 2020 in Remote Sensing
Verified
Journal Article
Open access
PDF file
Jiabin PuJiabin Pu
Jiabin Pu
Kai YanKai Yan
Kai Yan
GZGuohuan Zhou
Guohuan Zhou
YLYongqiao Lei
Yongqiao Lei
YZYingxin Zhu
Yingxin Zhu
RMRanga B. Myneni
Ranga B. Myneni
Uncertainty assessment of the moderate resolution imaging spectroradiometer (MODIS) leaf area index (LAI) and the fraction of photosynthetically active radiation absorbed by vegetation (FPAR) retrieval algorithm can provide a scientific basis for the usage and improvement of this widely-used product. Previous evaluations generally depended on the intercomparison with other datasets as well as direct validation using ground measurements, which mix the uncertainties from the model, inputs, and assessment method. In this study, we adopted the evaluation method based on three-dimensional radiative transfer model (3D RTM) simulations, which helps to separate model uncertainty and other factors. We used the well-validated 3D RTM LESS (large-scale remote sensing data and image simulation framework) for a grassland scene simulation and calculated bidirectional reflectance factors (BRFs) as inputs for the LAI/FPAR retrieval. The dependency between LAI/FPAR truth and model estimation serves as the algorithm uncertainty indicator. This paper analyzed the LAI/FPAR uncertainty caused by inherent model uncertainty, input uncertainty (BRF and biome classification), clumping effect, and scale dependency. We found that the uncertainties of different algorithm paths vary greatly (−6.61% and +84.85% bias for main and backup algorithm, respectively) and the “hotspot” geometry results in greatest retrieval uncertainty. For the input uncertainty, the BRF of the near-infrared (NIR) band has greater impacts than that of the red band, and the biome misclassification also leads to nonnegligible LAI/FPAR bias. Moreover, the clumping effect leads to a significant LAI underestimation (−0.846 and −0.525 LAI difference for two clumping types), but the scale dependency (pixel size ranges from 100 m to 1000 m) has little impact on LAI/FPAR uncertainty. Overall, this study provides a new perspective on the evaluation of LAI/FPAR retrieval algorithms.

30 October 2020 in IEEE Access
Verified
Journal Article
Open access
Jingxin YUJingxin YU
Jingxin YU
STSong Tang
Song Tang
LZLili Zhangzhong
Lili Zhangzhong
WZWengang Zheng
Wengang Zheng
LWLong Wang
Long Wang
Linlin XuLinlin Xu
Linlin Xu
Advance knowledge of soil water content (SWC) in the soil wetting layer of crop irrigation can help develop more reasonable irrigation plans and improve the efficiency of agricultural irrigation water use. To improve the accuracy of predicting SWC at multiple depths, the ResBiLSTM model was proposed, in which continuous meteorological and SWC data were gridded and transformed as model inputs, and then high-dimensional spatial and time series features were extracted by ResNet and BiLSTM, respectively, and integrated by a meta-learner. Meteorological, SWC and growth stage records data from seven typical maize monitoring stations in Hebei Province, China, during the 2016–2018 summer maize planting process were utilized for the training, evaluation and testing of the ResBiLSTM model, with model prediction targets set at 20cm, 30cm, 40cm and 50cm depths. Experimental results showed that: 1) ResBiLSTM model could achieve better model fit and prediction of meteorological and SWC data at all growth stages, with R 2 within [0.818, 0.991], average MAE within [0.79%, 2.00%], and the overall prediction accuracy ranked as follows: anthesis maturity stage > seedling stage > tassel stage; 2) The average MSE of the ResBiLSTM model for the prediction of SWC in the next 1–6 days was within [3.91%, 15.82%], and the prediction accuracy decreased with the extension of the prediction time; 3) Compared with the classical machine learning model and related deep learning models, the ResBiLSTM model was able to obtain better prediction accuracy performance.

30 October 2017 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
FLFan Li
Fan Li
DCDavid Clausi
David Clausi
Linlin XuLinlin Xu
Linlin Xu
AWAlexander Wong
Alexander Wong
The problem of limited labeled training samples is challenging for the classification of remote sensing imagery. We develop a joint classification and segmentation algorithm to address this problem. Our algorithm combines semisupervised learning and conditional random fields (CRFs) into a single framework. The multimodal Gaussian maximum-likelihood classifier is used to estimate the probabilities for the unary potentials of the CRF. Unlike traditional methods based on random fields, region merging is concatenated with the CRF inference to reduce the number of nodes iteratively. Moreover, a semisupervised technique called self-training is used, which iteratively enlarges the training sample set and retrains the classifier. The selection of training samples is based on the region information, so that the risk of assigning wrong labels is largely reduced. The proposed algorithm is applied to hyperspectral image classification, and results on benchmark data sets show that the proposed algorithm significantly improves classification performance after using self-training, and outperforms state-of-the-art spectral-spatial methods for limited labeled training samples.

23 January 2022 in Remote Sensing
Verified
Journal Article
Open access
PDF file
YLYoufeng Liu
Youfeng Liu
HYHonglei Yang
Honglei Yang
SWShizheng Wang
Shizheng Wang
Linlin XuLinlin Xu
Linlin Xu
JPJunhuan Peng
Junhuan Peng
The Woda area in the upper Jinsha River has steep terrain and broken structures, causing landslide disasters frequently. Here, we used the distributed scatterer interferometric SAR (DS-InSAR) method to monitor and analyze the Woda landslide area. With the DS-InSAR method, we derived the deformation of the Woda landslide area from 106 Sentinel-1A ascending images acquired between 5 November 2014 and 4 September 2019 and 102 Sentinel-1A descending images acquired between 31 October 2014 and 11 September 2019. The obvious advantage of the DS-InSAR method compared to the persistent scatterer (PS) InSAR (PS-InSAR) method is that the densities of the monitoring points were increased by 25.1% and 22.9% in the ascending and descending images, respectively. The two-dimensional deformation of the landslide area shows that the maximum surface deformation rate in the normal direction was −80 mm/yr, and in the east–west direction, 118 mm/yr. According to the rescaled range (R/S) analysis, the Hurst index values of the deformation trends were all greater than 0.5, which means the deformation trend will continue for some time. In addition, we analyzed the influencing factors and the deformation mechanism of the Woda landslide area and found that the surface deformation is closely related to the geological structure and precipitation, among which precipitation is the main factor triggering the deformation. Our monitoring results will help the local government to conduct regular inspections and strengthen landslide disaster prevention in low-coherence mountainous areas.