In the past two decades, unmanned aerial vehicles (UAVs) have been used in many scientific research fields for various applications. In particular, the use of UAVs for magnetic surveys has become a hot spot and is expected to be actively applied in the future. A considerable amount of literature has been published on the use of UAVs for magnetic surveys, however, how to choose the platform and reduce the interference of UAV to the collected data have not been discussed systematically. There are two primary aims of this study: (1) To ascertain the basis of UAV platform selection and (2) to investigate the characteristics and suppression methods of UAV magnetic interference. Systematic reviews were performed to summarize the results of 70 academic studies (from 2005 to 2021) and outline the research tendencies for applying UAVs in magnetic surveys. This study found that multi-rotor UAVs have become the most widely used type of UAVs in recent years because of their advantages such as easiness to operate, low cost, and the ability of flying at a very low altitude, despite their late appearance. With the improvement of the payload capacity of UAVs, to use multiple magnetometers becomes popular since it can provide more abundant information. In addition, this study also found that the most commonly used method to reduce the effects of the UAV’s magnetic interference is to increase the distance between the sensors and the UAV, although this method will bring about other problems, e.g., the directional and positional errors of sensors caused by erratic movements, the increased risk of impact to the magnetometers. The pros and cons of different types of UAV, magnetic interference characteristics and suppression methods based on traditional aeromagnetic compensation and other methods are discussed in detail. This study contributes to the classification of current UAV applications as well as the data processing methods in magnetic surveys.

Great progress has been made in the integration of Unmanned Aerial Vehicle (UAV) magnetic measurement systems, but the interpretation of UAV magnetic data is facing serious challenges. This paper presents a complete workflow for the detection of the subsurface objects, like Unexploded Ordnance (UXO), by the UAV-borne magnetic survey. The elimination of interference field generated by the drone and an improved Euler deconvolution are emphasized. The quality of UAV magnetic data is limited by the UAV interference field. A compensation method based on the signal correlation is proposed to remove the UAV interference field, which lays the foundation for the subsequent interpretation of UAV magnetic data. An improved Euler deconvolution is developed to estimate the location of underground targets automatically, which is the combination of YOLOv3 (You Only Look Once version 3) and Euler deconvolution. YOLOv3 is a deep convolutional neural network (DCNN)-based image and video detector and it is applied in the context of magnetic survey for the first time, replacing the traditional sliding window. The improved algorithm is more satisfactory for the large-scale UAV-borne magnetic survey because of the simpler and faster workflow, compared with the traditional sliding window (SW)-based Euler method. The field test is conducted and the experimental results show that all procedures in the designed routine is reasonable and effective. The UAV interference field is suppressed significantly with root mean square error 0.5391 nT and the improved Euler deconvolution outperforms the SW Euler deconvolution in terms of positioning accuracy and reducing false targets.

The thickness estimation of the top surface layer and surface layer, as well as the detection of road defects, are of great importance to the quality conditions of asphalt pavement. Although ground penetrating radar (GPR) methods have been widely used in non-destructive detection of pavements, the thickness estimation of the thin top surface layer is still a difficult problem due to the limitations of GPR resolution and the similar permittivity of asphalt sub-layers. Besides, the detection of some road defects, including inadequate compaction and delamination at interfaces, require further practical study. In this paper, a newly-developed vehicle-mounted GPR detection system is introduced. We used a horizontal high-pass filter and a modified layer localization method to extract the underground layers. Besides, according to lab experiments and simulation analysis, we proposed theoretical methods for detecting the degree of compaction and delamination at the interface, respectively. Moreover, a field test was carried out and the estimated results showed a satisfactory accuracy of the system and methods.

With the development of deep neural networks, speech enhancement technology has been vastly improved. However, commonly used speech enhancement approaches cannot fully leverage contextual information from different scales, which limits performance improvement. To address this problem, we propose a nested U-Net with self-attention and dense connectivity (SADNUNet) for monaural speech enhancement in the time domain. SADNUNet is an encoder-decoder structure with skip connections. In SADNUNet, the multi-scale aggregation block is proposed to explore more contextual information from different scales. By this means, the advantage of global and local speech features can be fully utilized to improve speech reconstruction ability. Furthermore, dense connectivity and self-attention are incorporated in the network for better feature extraction and utterance level context aggregation. The experimental results demonstrate that the proposed approach achieves on-par or better performance than other models in objective speech intelligibility and quality scores.

The Scalar Triangulation and Ranging (STAR) method, which is based upon the unique properties of magnetic gradient contraction, is a high real-time ferromagnetic target localization method. Only one measurement point is required in the STAR method and it is not sensitive to changes in sensing platform orientation. However, the localization accuracy of the method is limited by the asphericity errors and the inaccurate value of position leads to larger errors in the estimation of magnetic moment. To improve the localization accuracy, a modified STAR method is proposed. In the proposed method, the asphericity errors of the traditional STAR method are compensated with an iterative algorithm. The proposed method has a fast convergence rate which meets the requirement of high real-time localization. Simulations and field experiments have been done to evaluate the performance of the proposed method. The results indicate that target parameters estimated by the modified STAR method are more accurate than the traditional STAR method.

The measurement accuracy of the magnetometer system is closely related to the calibration effect. Available magnetometer calibration methods face the limitation on the assumption that the background magnetic field is uniform. The fact is that the experiment magnetic environment is inhomogeneous for the presence of human activities mainly. This problem could be partially addressed by an auxiliary large-scale triaxial Helmholtz coil to compensate the nonuniformity of magnetic field. This paper proposes a novel calibration method based on the tensor invariants in the nonuniform magnetic field without extra device. The inhomogeneity of background field implies a nonzero gradient field; a new correction model in the gradient field has been established. Calibration parameters are estimated to compensate the observed data using the concept of magnetic gradient tensor and the corresponding rotational tensor invariants, combined with the Levenberg Marquardt optimization. And after the calibration operation, the tensor invariants should be close to a constant. Numerical simulations have been designed and carried out, showing that the proposed algorithm had a good compensation performance with fast convergence, robustness in any gradient field. Field experiments with fluxgate magnetometer array were also conducted to verify this calibration procedure. The measured results are consistent with the simulation results. The root-mean-square (RMS) error is reduced from 616.239 to 1.608 nT/m, after calibration. In addition, the machining error of the frame, the time stability of the background field, the scale of gradient field, and robustness of this proposed method have been discussed.