Power Line Extraction and Reconstruction Methods from Laser Scanning Data: A Literature Review

Abstract

Electricity has become an indispensable source of energy, and power lines play a crucial role in the functioning of modern societies. It is essential to inspect power lines promptly and precisely in order to ensure the safe and secure delivery of electricity. In steep and mountainous terrain, traditional surveying methods cannot inspect power lines precisely due to their nature. Remote sensing platforms, such as satellite and aerial images, thermal images, and light detection and ranging (LiDAR) points, were utilised for the detection and inspection of power lines. Nevertheless, with the advancements of remote sensing technologies, in recent years, LiDAR surveying has been favoured for power line corridor (PLC) inspection due to active and weather-independent nature of laser scanning. Laser ranging data and the precise location of the LiDAR can be used to generate a three-dimensional (3D) image of the PLC. The resulting 3D point cloud enables accurate extraction of power lines and measurement of their distances from the forest below. In the literature, there have been many proposals for power line extraction and reconstruction for PLC modelling. This article examines the pros and cons of each domain method, providing researchers involved in three-dimensional modelling of power lines using innovative LiDAR scanning systems with useful guidelines. To achieve these objectives, research papers were analysed, focusing primarily on geoscience-related journals and conferences for the extraction and reconstruction of power lines. There has been a growing interest in examining the extraction and reconstruction of power line spans with single and multi-conductor configurations using different image and point-based techniques. Our study provides a comprehensive overview of the methodologies offered by various approaches using laser scanning data from the perspective of power line extraction applications, as well as to discuss the benefits and drawbacks of each approach. The comparison revealed that, despite the tremendous potential of aerial and mobile laser scanning systems, human intervention and post-processing actions are still required to achieve the desired results. In addition, the majority of the methods have been evaluated on the small datasets, and very few methods have been focused on multi-conductor extraction and reconstruction for power lines modelling. These barriers hinder the automated extraction and reconstruction of power line using LiDAR data and point to unexplored areas for further research and serve as useful guidelines for future research directions. Several promising directions for future LiDAR experiments using deep learning methods are outlined in the hope that they will pave the way for applications of PLC modelling and assessment at a finer scale and on a larger scale.

Power Line Extraction and Tree Risk Detection Based on Airborne LiDAR

“…Due to the extremely similar morphological features of conductors and ground wires, existing studies usually regard them as being in the same category [13][14][15][16]55]. However, the relevant departments of the State Grid Corporation of China make clear distinctions about both, so it is necessary to explore the automatic differentiation methods of conductors and ground wires.…”

“…Point cloud classification focuses on power line extraction, which is the basis for realizing the automatic detection of the safe distance between power lines and other ground objects [13][14][15][16], and is also the key research direction of many authors. In general, transmission line corridors contain multiple ground-object classes, such as terrain, vegetation, power lines, noise, roadway, and possibly a small number of buildings, which could make power line extraction more difficult.…”

The Effectiveness of a UAV-Based LiDAR Survey to Develop Digital Terrain Models and Topographic Texture Analyses

“…These can form the basis of various types of geomorphometric analyses [26], landforms measurements and land surface change analyses [27]. LiDAR-derived DTMs are currently used in many scientific fields [28,29] and industrialtechnical activities [30][31][32]. They provide information that is crucial or even necessary for environmental management in the broadest sense [33].…”

IPCONV: Convolution with Multiple Different Kernels for Point Cloud Semantic Segmentation

“…Point clouds provide precise 3D geometric information. They are used in digital twin cities [1,2], railways [3], cultural heritage [4], power line extraction and reconstruction [5], and autonomous driving [6]. However, they lack semantic information, hindering machine comprehension and usage.…”