Automatic Detection of UAV GCP Targets Using Line-Based Approach

With the advent and development of UAV technologies, UAV images are widely used in various fields since UAV photogrammetry has many advantages in terms of cost and accessibility. In addition, UAV photogrammetry has the advantage of enabling precise 3D surveying because it acquires images of higher spatial resolution with higher overlap compared to traditional aerial photogrammetry. UAV photogrammetry requires ground control points (GCPs) that are dense and evenly distributed throughout the study area. GCP surveying is generally conducted on-site, unlike automated UAV flight and image acquisition, which is a primary factor hindering time and labor cost reduction. In addition, pre-processing, such as UAV orthophoto, point cloud data, and digital elevation model (DEM) production, is performed automatically according to designated parameters, whereas matching GCP survey information with the images involves the intervention of an analyst. Therefore, in this study, the automatic extraction of UAV GCP targets and their centroids was investigated to increase the utilization of UAV photogrammetry and reduce the cost. Sequential steps of image thresholding, boundary detection, and buffered labeling detected a candidate area where ground targets exist. Then, the Hough transform was applied to the target candidates to extract two dominant lines and their intersection point representing the target center. The proposed method extracts the GCP targets from the images with high accuracy, and it was confirmed that it could be applied to complex urban areas. In addition, the GCP targets and their centroid points were successfully extracted from various land covers.