To improve the accuracy of the geographic positioning of a single aerial remote sensing image, the height information of a building in the image must be considered. Oblique remote sensing images are essentially two-dimensional images and produce a large positioning error if a traditional positioning algorithm is used to locate the building directly. To address this problem, this study uses a convolutional neural network to automatically detect the location of buildings in remote sensing images. Moreover, it optimizes an automatic building recognition algorithm for oblique aerial remote sensing images based on You Only Look Once V4 (YOLO V4). This study also proposes a positioning algorithm for the building target, which uses the imaging angle to estimate the height of a building, and combines the spatial coordinate transformation matrix to calculate high-accuracy geo-location of target buildings. Simulation analysis shows that the traditional positioning algorithm inevitably leads to large errors in the positioning of building targets. When the target height is 50 m and the imaging angle is 70°, the positioning error is 114.89 m. Flight tests show that the algorithm established in this study can improve the positioning accuracy of building targets by approximately 20%–50% depending on the difference in target height.

Target geolocation in long-range oblique photography (LOROP) is a challenging study due to the fact that measurement errors become more evident with increasing shooting distance, significantly affecting the calculation results. This paper introduces a novel high-accuracy target geolocation method based on multi-view observations. Unlike the usual target geolocation methods, which heavily depend on the accuracy of GNSS (Global Navigation Satellite System) and INS (Inertial Navigation System), the proposed method overcomes these limitations and demonstrates an enhanced effectiveness by utilizing multiple aerial images captured at different locations without any additional supplementary information. In order to achieve this goal, camera optimization is performed to minimize the errors measured by GNSS and INS sensors. We first use feature matching between the images to acquire the matched keypoints, which determines the pixel coordinates of the landmarks in different images. A map-building process is then performed to obtain the spatial positions of these landmarks. With the initial guesses of landmarks, bundle adjustment is used to optimize the camera parameters and the spatial positions of the landmarks. After the camera optimization, a geolocation method based on line-of-sight (LOS) is used to calculate the target geolocation based on the optimized camera parameters. The proposed method is validated through simulation and an experiment utilizing unmanned aerial vehicle (UAV) images, demonstrating its efficiency, robustness, and ability to achieve high-accuracy target geolocation.

In recent years, applications such as marine search and rescue, border patrol, etc. require electro-optical equipment to have both high resolution and precise geographic positioning abilities. The step and stare working based on a composite control system is a preferred solution. This paper proposed a step and stare system composed of two single-axis fast steering mirrors and a two-axis gimbal. The fast steering mirrors (FSMs) realize image motion compensation and the gimbal completes pointing control. The working principle and the working mode of the system are described first. According to the imaging optical path, the algorithm and control flow of the line of sight (LOS) and image motion compensation are developed. The proposed method is verified through ground imaging and flight tests. Under the condition of flight, the pointing accuracy of the target can be controlled within 15 m. The proposed algorithm can achieve effective motion compensation and get high-resolution images. This achieves high resolution and accurate LOS simultaneously.

Aerial photoelectric imaging payloads have become an important means of reconnaissance and surveillance in recent years. However, aerial images are easily affected by external conditions and have unclear edges, which greatly reduces the accuracy of imaging target recognition. This paper proposes the M-YOLOv5 model, which uses a shallow feature layer. The RFBs module is introduced to improve the receptive field and detection effect of small targets. In the neck network part, the BiFPN structure is used to reuse the underlying features to integrate more features, and a CBAM attention mechanism is added to improve detection accuracy. The experimental results show that the detection effect of this method on the DroneVehicle dataset is better than that of the original network, with the precision rate increased by 2.8%, the recall rate increased by 16%, and the average precision increased by 2.3%. Considering the real-time problem of target detection, based on the improved model, the Clight-YOLOv5 model is proposed, by lightweighting the network structure and using the depth-separable convolution optimization module. After lightweighting, the number of model parameters is decreased by 71.3%, which provides a new idea for lightweight target detection and proves the model’s effectiveness in aviation scenarios.

Object detection is a basic and important task in the field of aerial image processing and has gained much attention in computer vision. However, previous aerial image object-detection approaches have insufficient use of scene semantic information between different regions of large-scale aerial images. In addition, complex background and scale changes make it difficult to improve detection accuracy. To address these issues, we propose a relationship representation network for object detection in aerial images (RelationRS): (1) Firstly, multi-scale features are fused and enhanced by a dual relationship module (DRM) with conditional convolution. The dual relationship module learns the potential relationship between features of different scales and learns the relationship between different scenes from different patches in a same iteration. In addition, the dual relationship module dynamically generates parameters to guide the fusion of multi-scale features. (2) Secondly, the bridging visual representations module (BVR) is introduced into the field of aerial images to improve the object detection effect in images with complex backgrounds. Experiments with a publicly available object detection dataset for aerial images demonstrate that the proposed RelationRS achieves a state-of-the-art detection performance.

Aerial cameras are susceptible to optical aberrations under complex operating conditions, resulting in spatially varying image degradation. This degradation essentially stems from the point spread function (PSF) varying continuously across both spatial and temporal dimensions, exhibiting non-stationary and non-uniform characteristics. Such variability significantly increases the difficulty of correction and impairs image quality restoration. To address this challenge, we propose an image quality enhancement algorithm targeting spatially variant aberrations. First, due to the dynamic nature of the PSF, which makes it difficult to measure and model, we introduce a spatially varying PSF model guided by optical priors from Seidel aberrations. By constraining the blind estimation of PSFs using Seidel coefficients, the model explicitly characterizes their spatial variation. Then, to effectively correct the spatially varying aberrations while balancing computational efficiency and algorithmic performance, we adopt a strategy that combines patch-wise deconvolution with a Plug-and-Play (PnP) prior. Specifically, patch-wise deconvolution is used for initial correction, followed by a pretrained PnP network as a regularization term to impose global consistency across the image. Iterative refinement between deconvolution and PnP regularization further enhances image quality. Experimental results demonstrate that the proposed method outperforms existing state-of-the-art approaches in both synthetic and real-world scenarios, reducing the required number of iterations from 8 to 3, and improving the Neural Image Assessment (NIMA) scores by 7.49% and the Hyper Image Quality Assessment (HyperIQA) scores by 14.15% for visible-light cameras, while achieving improvements of 29.58% and 17.53% respectively for infrared cameras.