The global ecological environment faces many challenges. Landsat thematic mapper time-series, digital elevation models, meteorology, soil types, net primary production data, socio-economic data, and auxiliary data were collected in order to construct a comprehensive evaluation system for ecological vulnerability (EV) using multi-source remote sensing data. EV was divided into five vulnerability levels: potential I, slight II, mild III, moderate IV, and severe V. Then, we analyzed and explored the spatio-temporal patterns and driving mechanisms of EV in the region over the past 20 years. Our research results showed that, from 2001 to 2019, the DRB was generally characterized as being in the severe vulnerability class, with higher upstream and downstream EV classes and a certain amount of reduction in the midstream EV classes. Moreover, EV in the DRB continues to decrease. The spatio-temporal EV patterns in the DRB were significantly influenced by the relative humidity, average annual temperature, and vegetation cover over the past 20 years. Our work can provide a basis for decision-making and technical support for ecosystem protection, ecological restoration, and ecological management in the DRB.

The rapid development of Earth observation technology has promoted the continuous accumulation of images in the field of remote sensing. However, a large number of remote sensing images still lack manual annotations of objects, which makes the strongly supervised deep learning object detection method not widely used, as it lacks generalization ability for unseen object categories. Considering the above problems, this study proposes a few-shot remote sensing image object detection method that integrates context dependencies and global features. The method can be used to fine-tune the model with a small number of sample annotations based on the model trained in the base class, as a way to enhance the detection capability of new object classes. The method proposed in this study consists of three main modules, namely, the meta-feature extractor (ME), reweighting module (RM), and feature fusion module (FFM). These three modules are respectively used to enhance the context dependencies of the query set features, improve the global features of the support set that contains annotations, and finally fuse the query set features and support set features. The baseline of the meta-feature extractor of the entire framework is based on the optimized YOLOv5 framework. The reweighting module of the support set feature extraction is based on a simple convolutional neural network (CNN) framework, and the foreground feature enhancement of the support sets was made in the preprocessing stage. This study achieved beneficial results in the two benchmark datasets NWPU VHR-10 and DIOR. Compared with the comparison methods, the proposed method achieved the best performance in the object detection of the base class and the novel class.

Unmanned aerial vehicles (UAVs) have become an important technology for acquiring high-resolution remote sensing images. Because most space optical imaging systems of UAVs work in environments affected by vibrations, the optical axis motion and image plane jitter caused by these vibrations easily result in blurring of UAV images. In the paper; we propose an advanced UAV image deblurring method based on a discriminative model comprising a classifier for blurred and sharp UAV images which is embedded into the maximum a posteriori framework as a regularization term that constantly optimizes ill-posed problem of blind image deblurring to obtain sharper UAV images. Compared with other methods, the results show that in image deblurring experiments using both simulated and real UAV images the proposed method delivers sharper images of various ground objects.