Assessing the number and placement of ground control points in low-cost UAV photogrammetry

Abstract

Listen

Translate article

This study examines the impact of the number and placement of Ground Control Points (GCPs) on the accuracy of photogrammetric results using low-cost UAVs. The relevance of this research stems from the growing popularity and applicability of UAV photogrammetry, particularly in areas with diverse terrains and requirements for precise spatial data. The experiment leverages modern photogrammetric techniques, including Structure from Motion (SfM) algorithms, to analyze how GCP configurations affect error distribution and model accuracy. The study's first phase focused on evaluating the influence of GCP placement at varying heights. A photogrammetric survey was conducted at the Kyiv Hippodrome, utilizing terrestrial laser scanning (TLS) to establish high-precision coordinates for control points. The data acquisition involved the DJI Phantom 4 Pro V2 UAV, with multiple flight missions capturing images at 30-degree camera deviations from the nadir. Points were systematically analyzed by alternating their roles as GCPs and Control Points (CPs). The results demonstrated that errors significantly increased when CPs were located further from the UAV camera, emphasizing the need for proximity in GCP placement. The second phase analyzed how GCP configurations and quantities influence photogrammetric model accuracy. By forming 12 groups of GCPs, each varying in distribution and number, the study identified optimal setups for minimizing errors. Groups with evenly distributed points across the survey area, comprising at least eight GCPs, exhibited the lowest root mean square errors (RMSE). Conversely, configurations with GCPs concentrated along a single side or solely on the survey area's edges resulted in substantial inaccuracies. Key findings reveal that an effective GCP placement strategy involves prioritizing proximity to the UAV camera and achieving even distribution across the surveyed area. Additionally, configurations with fewer than eight GCPs tend to suffer from sharp declines in accuracy. The research underscores the importance of balancing GCP quantity and placement for achieving reliable photogrammetric outputs in low-cost UAV applications