Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zhi-Qiang Liu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Enhancing Satellite Image Sequences through Multi-Scale Optical Flow-Intermediate Feature Joint Network

    by Keli Shi, Zhi-Qiang Liu, Weixiong Zhang, Ping Tang, Zheng Zhang

    Satellite time-series data contain information in three dimensions—spatial, spectral, and temporal—and are widely used for monitoring, simulating, and evaluating Earth activities. However, some time-phase images in the satellite time series data are missing due to satellite sensor malfunction or adverse atmospheric conditions, which prevents the effective use of the data. Therefore, we need to complement the satellite time series data with sequence image interpolation. Linear interpolation methods and deep learning methods that have been applied to sequence image interpolation lead to large errors between the interpolation results and the real images due to the lack of accurate estimation of pixel positions and the capture of changes in objects. Inspired by video frame interpolation, we combine optical flow estimation and deep learning and propose a method named Multi-Scale Optical Flow-Intermediate Feature Joint Network. This method learns pixel occlusion and detailed compensation information for each channel and jointly refines optical flow and intermediate features at different scales through an end-to-end network together. In addition, we set a spectral loss function to optimize the network’s learning of the spectral features of satellite images. We have created a time-series dataset using Landsat-8 satellite data and Sentinel-2 satellite data and then conducted experiments on this dataset. Through visual and quantitative evaluation of the experimental results, we discovered that the interpolation results of our method retain better spectral and spatial consistency with the real images, and that the results of our method on the test dataset have a 7.54% lower Root Mean Square Error than other approaches.


  • Open Access

    Bandpass Alignment from Sentinel-2 to Gaofen-1 ARD Products with UNet-Induced Tile-Adaptive Lookup Tables

    by Zhi-Qiang Liu, Zhao Wang, Zhitao Zhao, Lianzhi Huo, Ping Tang, Zheng Zhang

    The successful launching of more satellites in recent years has made data fusion an important and promising task because it can significantly increase the temporal frequency of the resulting time series data. To improve the usability of Gaofen-1 analysis ready data (GF1-ARD), Sentinel-2 (S2) is selected to enhance the temporal resolution of GF1-ARD due to their similar characteristics and short revisit period. Before constructing a denser time series from different platforms, bandpass alignment is required. Most researchers implement bandpass alignment using the linear model. However, the transformed bands of S2 by the linear model cannot match GF1-ARD well due to the limited globally shared parameters. In contrast, local-derived lookup tables (LUTs) can better address this problem. Inspired by the powerful capability of deep learning, we develop a model based on the U-shaped network (UNet) to learn tile-adaptive LUTs. Specifically, the LUTs are adaptively learned from the histogram of the S2 tile. Given that the bandpass alignment can be viewed as a histogram matching process, the expected LUTs are believed to be highly correlated with the input histogram. In addition, a simple convolutional module is further introduced to address the pixel-level misregistration. We have created a large-scale dataset and conducted extensive experiments on it to evaluate the competitive performance of the proposed model. Meanwhile, extensive visualizations are generated to illustrate the mechanism of our model. Furthermore, the temporal frequency of S2 and GF1-ARD is thoroughly assessed to demonstrate that bandpass alignment can significantly improve the temporal resolution of GF1-ARD.


  • Open Access

    Spectral Mixing Theory-Based Double-Branch Network for Spectral Super-Resolution

    by Lingyu Sha, Wenjuan Zhang, Bing Zhang, Zhiqiang Liu, Zhen Li

    Due to the limitations of imaging systems, the swath width of available hyperspectral images (HSIs) is relatively narrow, making it difficult to meet the demands of various applications. Therefore, an appealing idea is to expand the width of HSIs by using widely covered multispectral images (MSIs), called spectral super-resolution (SSR) of MSIs. According to the radiation transmission process of the imaging system, the spectral mixing characteristics of ground objects can be described by the linear spectral mixing model (LSMM). Inspired by the linear mixed part and nonlinear residual part of the LSMM, we propose a double-branch SSR network. To generate wide HSIs, a spectral mixing branch is designed to extract abundances from wide MSIs and adaptively learn hyperspectral endmembers from narrow HSIs. Furthermore, considering the nonlinear factors in the imaging system and atmospheric transmission, a nonlinear residual branch is built to complement the spectral and spatial details. Finally, the SSR result can be obtained with the fusion of linear and nonlinear features. To make the network structure achieve corresponding physical significance, we constrain the network through joint loss functions at different stages. In addition to two simulated datasets with limited coverage, our model is also evaluated on a real MSI–HSI dataset in a larger area. Extensive experiments show the superiority of the proposed model compared with state-of-the-art baselines. Moreover, we visualized the internal results of our network and conducted ablation experiments on a single branch to further demonstrate its effectiveness. In the end, the influence of network hyperparameters, including endmembers and loss function weight coefficient, is discussed and analyzed in detail.


  • Open Access

    CNN-Enhanced Heterogeneous Graph Convolutional Network: Inferring Land Use from Land Cover with a Case Study of Park Segmentation

    by Zhi-Qiang Liu, Ping Tang, Weixiong Zhang, Zheng Zhang

    Land use segmentation is a fundamental yet challenging task in remote sensing. Most current methods mainly take images as input and sometimes cannot achieve satisfactory results due to limited information. Inspired by the inherent relations between land cover and land use, we investigate land use segmentation using additional land cover data. The topological relations among land cover objects are beneficial for bridging the semantic gap between land cover and land use. Specifically, these relations are usually depicted by a geo-object-based graph structure. Deep convolutional neural networks (CNNs) are capable of extracting local patterns but fail to efficiently explore topological relations. In contrast, contextual relations among objects can be easily captured by graph convolutional networks (GCNs). In this study, we integrated CNNs and GCNs and proposed the CNN-enhanced HEterogeneous Graph Convolutional Network (CHeGCN) to incorporate local spectral-spatial features and long-range dependencies. We represent topological relations by heterogeneous graphs which are constructed with images and land cover data. Afterwards, we employed GCNs to build topological relations by graph reasoning. Finally, we fused CNN and GCN features to accomplish the inference from land cover to land use. Compared with other homogeneous graph-based models, the land cover data provide more sufficient information for graph reasoning. The proposed method can achieve the transformation from land cover to land use. Extensive experiments showed the competitive performance of CHeGCN and demonstrated the positive effects of land cover data. On the IoU metric over two datasets, CHeGCN outperforms CNNs and GCNs by nearly 3.5% and 5%, respectively. In contrast to homogeneous graphs, heterogeneous graphs have an IoU improvement of approximately 2.5% in the ablation experiments. Furthermore, the generated visualizations help explore the underlying mechanism of CHeGCN. It is worth noting that CHeGCN can be easily degenerated to scenarios where no land cover information is available and achieves satisfactory performance.


  • Open Access

    Seeded Classification of Satellite Image Time Series with Lower-Bounded Dynamic Time Warping

    by Zheng Zhang, Ping Tang, Changmiao Hu, Zhiqiang Liu, Weixiong Zhang, Liang Tang

    Satellite Image Time Series (SITS) record the continuous temporal behavior of land cover types and thus provide a new perspective for finer-grained land cover classification compared with the usual spectral and spatial information contained in a static image. In addition, SITS data is becoming more accessible in recent years due to newly launched satellites and accumulated historical data. However, the lack of labeled training samples limits the exploration of SITS data, especially with sophisticated methods. Even with a straightforward classifier, such as k-nearest neighbor, the accuracy and efficiency of the SITS similarity measure is also a pending problem. In this paper, we propose SKNN-LB-DTW, a seeded SITS classification method based on lower-bounded Dynamic Time Warping (DTW). The word “seeded” indicates that only a few labeled samples are required, and this is not only because of the lack of labeled samples but also because of our aim to explore the rich information contained in SITS, rather than letting training samples dominate the classification results. We use a combination of cascading lower bounds and early abandoning of DTW as an accurate yet efficient similarity measure for large scale tasks. The experimental results on two real SITS datasets demonstrate the utility of the proposed SKNN-LB-DTW, which could become an effective solution for SITS classification when the amount of unlabeled SITS data far exceeds the labeled data.