Junwu Dong

As a fundamental task in remote sensing interpretation, semantic segmentation of remote sensing images intends to allocate a definite class to each pixel in the image. Fast and efficient semantic segmentation of high-resolution remote sensing images provides helps to capture the real surface covering and plays an essential role in urban planning and dynamic monitoring. However, there are still some limitations in the previous remote sensing image semantic segmentation model for urban scenes, such as the low weight of small target pixels and the tiny target size leading to the unsatisfactory recognition and segmentation results of the model for small target features. Meanwhile, the deeper and broader feature extraction module in the semantic segmentation network usually leads to more redundant parameters, which takes a lot of computation time. Thus, we propose a lightweight semantic segmentation network based on knowledge distillation combined with a multiscale pyramidal pooling mod-ule and attention mechanism named KDMSANet, which enhanced the ability to fuse and focus on shallow features. Then, we trained teacher-student models to obtain lightweight network models through a model pruning and distillation framework. Experiments on Vaihingen and Potsdam datasets demonstrated that the network we designed significantly reduces the number of parameters while ensuring almost constant accuracy. Compared to the pre-compression model, the student model reduced in size by 43.6% the training efficiency was improved by 22.3%, while the accuracy reached 99.30% of the teacher model

Accurate canopy and single-tree mapping is important to obtain information on the ecological structure and biogeophysical parameters for forests. Although some airborne radars can retrieve canopy and single-tree information within a smaller area, the optical satellite imagery-based approaches for rapidly and accurately mapping them over a large region are still limited. In this study, based on Eucalyptus canopy and single-tree texture and spectral features, we proposed a mapping approach using the combinations of image morphology, the Otsu method, and an adaptive iterative erosion algorithm (EUMAP). Then, we applied the commonly used red/green/blue bands from the high-resolution satellite images, which are freely available, to map the canopy and single-tree in Eucalyptus plantations in southern China. EUMAP consists of two steps: (i) Eucalyptus canopy identification for various canopy density regions; (ii) adaptive iterative erosion to separate single-tree. Our study was conducted in the Chengzhong and Liubei districts of Liuzhou city, China. The accuracy evaluation was carried out in the state-owned Sanmenjiang Forest Farm. The results showed that the average F1 score for mapping canopy and single-tree reached 88.34% and 86.40%, respectively. For the whole study area, there were 7033021 Eucalyptus trees and the average density was 819 trees per hectare. The approach adopted in this study, combining the prior knowledges about image morphology and single-tree texture features of Eucalyptus plantations, was highly efficient for satellite image processing and had excellent applicability to large-scale Eucalyptus plantations mapping. Our study highlights the necessary of prior knowledges for forest mapping using satellite images without requiring a training sample and provides a universal approach of accurately large-scale mapping for specific forest species with common red/green/blue images.

Considering the United Nations’ Sustainable Development Goals (SDGs) and the need for a balanced spatial distribution of urban medical resources capable of perspective of hierarchical diagnosis and treatment, i.e. providing continuous and accessible medical services during potential public health emergencies, we assessed accessibility and service capacity of the three hospital levels in Beijing. Using geographical information systems (GIS) and the two-step floating catchment area method with the street as research unit, we found that there is an over-supply of medical resources in the centre of the city with weaker support in the peripheral areas as manifested by less supply in relation to popular demand of medical services. The spatial distribution of hospitals at all levels and their resources was found to be uneven: 82.4% of the residents can reach a tertiary hospital (a hospital offering advanced specialized medical and health services to multiple regions) within a 15-minute drive; 50.6% can reach a secondary hospital (a hospital offering comprehensive medical and health services to various communities) within a 10-minute drive; and 77.6% can reach a primary hospital (a hospital directly delivering prevention, medical treatment, healthcare, and rehabilitation services to the community of a certain population) within a 15- minute walk. It was noted that the supply/demand balance of medical resources in the tertiary hospitals decreases from the centre to the periphery, while the secondary hospitals show a dual-centre pattern and the primary hospitals a more uneven distribution, with oversupply in the East and the opposite in the Centre. The results of the study provide supplementary decision support for improving the hierarchical diagnosis and treatment system and accelerate the overall deployment of medical resources.

SDG15.3 aims to achieve “Land Degradation Neutrality (LDN)”, but its only indicator (SDG15.3.1) is designed for national-level assessment and monitoring, and is not suitable for local-level applications. Thus, taking Inner Mongolia as the study area, this paper provides a localized reform of SDG15.3.1 based on the local context, and assesses the progress of SDG15.3 in the study area (2000-2020) at multiple levels (indicator, specific, and overall). The Moran’I and Standard Deviation Ellipse (SDE) are also utilized to analyze the spatial–temporal change of land degradation. The results show that as of 2020, the proportion of land degradation and improvement areas to the total area was 7.51% and 9.42%, respectively. Inner Mongolia had generally met the goal of SDG15.3, but on the indicator level, water erosion still falls far below the standard of SDG15.3. Additionally, at the spatial level, 3 out of 12 municipalities and 71 out of 103 counties had not achieved LDN, with a pattern of low LDN levels in the southeast and high in the northwest at the county scale. This indicates that the progress of SDG15.3 is extremely uneven both at the indicator and spatial levels. Thus, it is essential to continue to promote land degradation management in Inner Mongolia to achieve LDN in the entire area and across all aspects.

High-resolution remote-sensing images cover more feature information, including texture, structure, shape, and other geometric details, while the relationships among target features are more complex. These factors make it more complicated for classical convolutional neural networks to obtain ideal results when performing a feature classification on remote-sensing images. To address this issue, we proposed an attention-based multiscale max-pooling dense network (DMAU-Net), which is based on U-Net for ground object classification. The network is designed with an integrated max-pooling module that incorporates dense connections in the encoder part to enhance the quality of the feature map, and thus improve the feature-extraction capability of the network. Equally, in the decoding, we introduce the Efficient Channel Attention (ECA) module, which can strengthen the effective features and suppress the irrelevant information. To validate the ground object classification performance of the multi-pooling integration network proposed in this paper, we conducted experiments on the Vaihingen and Potsdam datasets provided by the International Society for Photogrammetry and Remote Sensing (ISPRS). We compared DMAU-Net with other mainstream semantic segmentation models. The experimental results show that the DMAU-Net proposed in this paper effectively improves the accuracy of the feature classification of high-resolution remote-sensing images. The feature boundaries obtained by DMAU-Net are clear and regionally complete, enhancing the ability to optimize the edges of features.

Habitat quality (HQ) is an important indicator to characterize the level of biodiversity and ecosystem services, and can reflect the quality of the human living environment. Changes in land use can disturb regional HQ. Current research mostly focuses on assessing the good or bad quality of regional habitats, and less on the spatial response relationship between land use change and HQ, and even fewer studies on finely distinguishing the impact of land use types on HQ. Therefore, taking Three Gorges Reservoir Area (TGRA) of China as the study area, this paper first analyzes the land use change of study area by using the land use transfer matrix, land use rate model and landscape pattern index, and then combines the InVEST model with the multi-scale geographically weighted regression (MGWR) model to build a refined assessment framework to quantitatively assess the spatial and temporal evolution patterns of HQ, and then analyse in detail the spatial response relationship of each land use type change on the impact of HQ. The results showed that from 2000 to 2020, the land use in the TGRA shows a changing state of “urban expansion, cultivated land shrinkage, forest land growth, and grassland degradation”. With the change in land use, the habitat quality index (HQI) in the study area showed an “ increase first and then decline” change characteristics, and the HQ degradation was more obvious in the areas with intense human activities. The impact of land use change over the past 20 years on HQ in the TGRA has significant spatial and temporal heterogeneity, with changes in paddy and dryland having mainly negative impacts on HQ, and changes in sparse land, shrubland, and medium-cover grassland having mainly positive impacts on HQ. This paper mainly provides a research framework for refined assessment, and the results can provide scientific support for land planning and ecological protection in the TGRA, and the research methods and ideas can provide references for similar research.

Complex traffic scenes are susceptible to the occlusion of moving targets and blurred target images, resulting in incomplete extraction of target trajectories within the field of view (FOV). To address this issue, we proposed a complete trajectory extraction algorithm for moving targets (TraEA) that considers the multi-level semantic features of traffic scenes following the FOV of single-camera surveillance videos. TraEA utilizes semantic information from the multi-level scene structure, such as image pixel features, geometric features, as well as temporal and spatial relationships between targets within each keyframe image. The semantic information helps extract and optimize the complete trajectory of a target, to assemble the target’s moving trajectory with improved integrity. An experiment with real-world traffic videos collected in Beijing, China validated the proposed algorithm and showed that TraEA improved the completeness, accuracy and robustness of target trajectory extraction. Traffic trajectories from the experiment can facilitate traffic supervision by analyzing the moving behaviors of individual targets. While the proposed TraEA algorithm is applicable to extract trajectories from videos in other domains, such as animal movements, potential privacy infringement and abusive applications warrant cautions of ethical considerations.