Abstract In this paper, the pixel-level corner position extracted by the Harris algorithm is used as the initial value of sub-pixel corner detection, and then it is used to detect the sub-pixel corner detection. The sub-pixel corner detection algorithm based on the inner product of the gray gradient has weak anti-noise and poor anti-distortion ability, and the sub-pixel corner detection algorithm based on Gauss quadric surface fitting has a high dependence on the initial value. In this case, a quadric fitting based on an iterative search thought algorithm of subpixel corner detection is proposed in this paper. The experiment shows that the improved algorithm has a stronger ability in anti-noise and anti-distortion and low dependence on the initial value. Using this algorithm to calibrate the camera, the re-projection error is smaller and the calibration accuracy is higher.

The application scope of UAV in the power field is gradually expanding, which puts forward higher requirements for the accurate landing of UAV. At this stage, the fixed-point landing of UAV mainly depends on RTK, but RTK is easy to be disturbed by the external environment, so it can not meet the requirements of fine landing. This paper proposes an improved yolov4-tiny algorithm, which adds the convolutional attention module to the yolov4-tiny network structure feature pyramid to reduce the interference of complex backgrounds on target recognition. In this system, the UAV is guided to fly above the nest by RTK in the first, and then the landing code is recognized by improved yolov4 tiny algorithm, so as to realize the accurate landing of UAV. Through the field fine landing test of UAV, it is found that the landing accuracy of the proposed technology is high and meets the landing requirements.

With the vigorous development of the solar photovoltaic industry, the operation and maintenance of photovoltaic power stations has ushered in huge challenges. To solve the problems of low efficiency, poor accuracy, and high cost of traditional manual inspection of photovoltaic power stations, the UAV inspection system for photovoltaic has become a new solution to solve the pain points of the industry, which can improve the inspection efficiency greatly. In photovoltaic inspection systems, defect detection is the most important goal. In this paper, a defect detection algorithm based on improved yolov5 for photovoltaic modules is proposed. The convolution block attention module (CBAM) is introduced to extract the attention area, and the weighted non-maximum suppression (NMS) is replaced by DIOU_NMS to improve the recognition accuracy of the model. The experimental results show that the algorithm can effectively identify the hot spots and diode defects of photovoltaic modules, which lays a solid foundation for the fully autonomous inspection of photovoltaic power stations.

With the continuous expansion of transmission line scale, line inspection has faced great challenges. To solve the problems of high labor intensity, low efficiency, and difficult working conditions in traditional manual inspection, autonomous inspection based on unmanned aerial vehicle (UAV) came into being. In this paper, an electrical component recognition algorithm based on improved yolov5s is proposed. Self-calibrated convolutions (SCConv) is introduced to improve the feature extraction capability. The Gamma transform is used to preprocess the image set, which can improve the robustness of the algorithm to overexposed and underexposed scenarios. The experimental results show that the proposed algorithm can be used for real-time electrical component recognition at the front end of UAV. The recognition accuracy of various typical components can reach more than 98%, which can significantly improve the acquisition quality of inspection images and lay a good foundation for further data processing at the back end of inspection system.

针对无人机载激光雷达系统应用中的数据传输问题,本文提出了一种基于4G的网络数据实时传输方案.在该方案中,结合无人机载激光雷达技术的工作原理,分别从硬件、软件、无线传输三个部分展开研究.在硬件部分,综合应用了Devkit8500D硬件开发平台和大疆M100无人机二次开发平台,并在此基础上进行了硬件改造;在软件部分,分别对嵌入式软件和地面端软件进行设计;在无线传输部分,以LTE作为核心网络,通过4G采集卡和Socket协议实现无人机端和地面端的通信,保证了激光雷达数据的采集.最后,将该系统用于电力巡检,最终得到了较为清晰的三维图像.

理想情况下,双目立体视觉模型是平行放置的,三维坐标易于解算.但在实际安装摄像机的时候,光轴无法做到平行,标定所得的重投影矩阵和左右图像匹配都存在误差,影响了三维坐标解算精度.针对此种情况,对非平行双目结构三维坐标解算进行了深入的研究.传统的最小二乘法和归一化最小二乘法都是利用求解联立的超定方程组得到最优解,将最优解看作是空间物点的三维坐标,但是它们都没有考虑所建立的超定方程组所代表的几何意义.从几何学角度出发,推导出一种基于异面直线公垂线中点的三维坐标解算公式,通过在无人机平台上测距实验,验证了此方法比最小二乘法和归一化最小二乘法具有更高的精度.