GUO ZH

Ms. GUO ZH


Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

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

Representative Publications

View All Publications
5 December 2019 in Remote Sensing
Verified
Journal Article
Open access
PDF file
WLWei Liu
Wei Liu
MYMengYuan Yang
MengYuan Yang
MXMeng Xie
Meng Xie
GZGUO ZH
GUO ZH
ELErzhu Li
Erzhu Li
DWDong Wang
Dong Wang
Accurate extraction of buildings using high spatial resolution imagery is essential to a wide range of urban applications. However, it is difficult to extract semantic features from a variety of complex scenes (e.g., suburban, urban and urban village areas) because various complex man-made objects usually appear heterogeneous with large intra-class and low inter-class variations. The automatic extraction of buildings is thus extremely challenging. The fully convolutional neural networks (FCNs) developed in recent years have performed well in the extraction of urban man-made objects due to their ability to learn state-of-the-art features and to label pixels end-to-end. One of the most successful FCNs used in building extraction is U-net. However, the commonly used skip connection and feature fusion refinement modules in U-net often ignore the problem of feature selection, and the ability to extract smaller buildings and refine building boundaries needs to be improved. In this paper, we propose a trainable chain fully convolutional neural network (CFCN), which fuses high spatial resolution unmanned aerial vehicle (UAV) images and the digital surface model (DSM) for building extraction. Multilevel features are obtained from the fusion data, and an improved U-net is used for the coarse extraction of the building. To solve the problem of incomplete extraction of building boundaries, a U-net network is introduced by chain, which is used for the introduction of a coarse building boundary constraint, hole filling, and "speckle" removal. Typical areas such as suburban, urban, and urban villages were selected for building extraction experiments. The results show that the CFCN achieved recall of 98.67%, 98.62%, and 99.52% and intersection over union (IoU) of 96.23%, 96.43%, and 95.76% in suburban, urban, and urban village areas, respectively. Considering the IoU in conjunction with the CFCN and U-net resulted in improvements of 6.61%, 5.31%, and 6.45% in suburban, urban, and urban village areas, respectively. The proposed method can extract buildings with higher accuracy and with clearer and more complete boundaries.

16 February 2022 in International Journal of Remote Sensing
Verified
Journal Article
Closed access
GZGUO ZH
GUO ZH
WLWei Liu
Wei Liu
JXJiawei Xu
Jiawei Xu
ELErzhu Li
Erzhu Li
XLXing Li
Xing Li
JZJiaxing Zhang
Jiaxing Zhang
Quality checks are important to ensure the accuracy and reliability of spatial data. However, current methods primarily focus on attributes and geometric information, and methods to verify the accuracy of land-type classification for vector patches are lacking. Therefore, in this paper, we proposed a framework for the automatic verification of the land type of vector patches, including the segmentation of complex vector patches, automatic acquisition of training samples, and automatic extraction of suspicious patches based on deep learning scene interpretation results. First, an innovative method for segmentation based on an oriented bounding box was proposed. Then, scene interpretation from the segmented processing units was performed. Finally, the scene interpretation results and the original data for vector patches were combined, and segmentation units that do not match the category information were automatically identified. The suspicious patches were extracted, and the land-type authenticity of vector patches was automatically checked. Experimental results showed that, in the study area, the accuracy, precision, recall, and F1 values for the model based on first-level land categories were 0.989, 0.842, 0.906, and 0.873, respectively, and those for second-level land categories were 0.994, 0.770, 0.891, and 0.826, respectively. Accordingly, the newly developed method provides reliable technical support for checking the land-type authenticity of vector patches.