Unmanned aerial vehicles (UAVs) have become important tools for power transmission line inspection. Cameras installed on the platforms can efficiently obtain images containing information about equipment. However, most of existing inspection systems cannot perform automatic real-time detection components. In this paper, an system incorporating UAV remote sensing with binocular visual perception technology is developed to accurately detect and locate equipment in real time. The consists a...

The existing unmanned aerial vehicle (UAV)-based electric transmission line inspection systems generally adopt manual control and follow the predefined path, which reduces efficiency makes a high cost. In this article, UAV system with advanced embedded processors binocular visual sensors is developed to generate guidance information from power lines in real-time achieve automatic inspection. To realize 3-D autonomous perception of lines, we first propose an end-to-end convolutional neural...

Node deployment is one of the fundamental tasks for underwater acoustic sensor networks (UASNs) where strategy supports many network services, such as topology control, routing, and boundary detection. Due to complex environment in three-dimensional (3D) space unique characteristics channel, factors need be considered specifically during UASNs. Thus, issues UASNs are significantly different from those wireless (WSNs). an attractive research topic upon which a large number algorithms have...

Sidescan sonar image segmentation is a very important issue in underwater object detection and recognition. In this paper, robust fast method for sidescan proposed, which deals with both speckle noise intensity inhomogeneity that may cause considerable difficulties segmentation. The proposed integrates the nonlocal means-based filtering (NLMSF), coarse using k -means clustering, fine an improved region-scalable fitting (RSF) model. NLMSF used before to effectively remove while preserving...

Most CNN models exhibit two major flaws in hyper-spectral image (HSI) restoration tasks. First, limited high-dimensional HSI training examples exacerbate the difficulty of deep learning methods effective spatial and spectral representations. Second, existing CNN-based model local relations present limitations capturing long-range dependencies. In this paper, we customize a novel dual-stream Transformer (DSTrans) for restoration, which mainly consists attention feed-forward network....

Real-time scene comprehension is the basis for automatic electric power inspection. However, existing RGB-based methods may achieve unsatisfied performance when dealing with complex scenarios, insufficient illumination or occluded appearances. To solve this problem, by cooperating visual and thermal images, Dual-Space Graph-based Interaction Network (DSGBINet) proposed to all-day time semantic segmentation of equipment in high-voltage transmission line transformer substation scenes....

As a typical deep-learning model, Convolutional Neural Networks (CNNs) can be exploited to automatically extract features from images using the hierarchical structure inspired by mammalian visual system. For image classification tasks, traditional CNN models employ softmax function for classification. However, owing limited capacity of function, there are some shortcomings in To deal with this problem, new method combining Biomimetic Pattern Recognition (BPR) CNNs is proposed BPR performs...

Binocular vision uses the parallax principle of human eye to obtain 3D information an object, which is widely used as important means acquiring for reconstruction tasks. To improve accuracy and efficiency reconstruction, we propose a method that combines second-order semiglobal matching, guided filtering Delaunay triangulation. First, existing matching improved, smoothness constraint multiple angle directions added cost generate more robust disparity map. Second, coordinates all points are...

In this paper, a measurement system based on internal cooperation of cameras in binocular vision is proposed. terms the hardware structure, composed module, controllable platform and an electronic gradienter, which can be used to collect images measure horizontal vertical angles between module objects. data processing, new method monocular (MBC) put forward. Firstly, depth values some effective points are obtained according feature point matching results. Then, propagation algorithm MBC...

Machine vision inspection technology provides an efficient tool for surface defects inspection. However, because of the multiformity defects, existing machine methods are limited by application scenarios. In order to improve versatility algorithms, and process various kinds images more accurately, we propose a new adaptive method defect detection, named neighborhood gray-level difference using multidirectional fluctuation. This changes thresholds step values extracting gray-level-fluctuating...

Thermal infrared imaging could detect hidden faults in various types of high-voltage power equipment, which is great significance for inspections. However, there are still certain issues with thermal-imaging-based abnormal heating detection methods due to varying appearances regions and complex temperature interference from backgrounds. To solve these problems, a contour-based instance segmentation network first proposed utilize thermal (T) visual (RGB) images, realizing high-accuracy...

The existing size measurement systems cannot meet the requirements of non-contact tasks, and main challenge is how to detect various objects improve accuracy. In order solve these problems, a novel system was proposed in this paper. system, three-camera model with variable baselines designed based on trinocular vision. Three binocular vision subsystems were composed three cameras, which used obtain depth information from different shooting angles, between cameras could be adjusted according...

In recent years, with the rapid development of video surveillance infrastructure, more and intelligent systems have employed computer vision pattern recognition techniques. this paper, we present a novel system used for management road vehicles based on Intelligent Visual Internet Things (IVIoT). The has ability to extract vehicle visual tags urban roads; in other words, it can label any by means therefore easily recognize tags. nodes designed be installed not only roads providing basic...

Human action recognition in videos is a topic of active research computer vision. Dense trajectory (DT) features were shown to be efficient for representing state-of-the-art approaches. In this paper, we present more effective approach video representation using improved salient dense trajectories: first, detecting the motion region and extracting trajectories by tracking interest points each spatial scale separately then refining via analysis saliency. Then, compute several descriptors...

Overhead transmission line corridor detection is important to ensure the safety of power facilities. Frequent and uncertain changes in environment requires an efficient autonomous UAV inspection system, whereas existing UAV-based systems has some shortcomings control model ground clearance measurement. For one thing, manual risk striking lines because it difficult for manipulators judge distance between fuselage accurately. another, methods based on usually depend LiDAR (Light Detection...