A UAV-Based Waterway Inspection Method Using Extended Kalman Filter and Multi-band Difference

Abstract. To address the limitations of traditional manual inspection methods for waterway regulation structures, this paper presents an unmanned aerial vehicle (UAV)-based inspection approach that integrates the Extended Kalman Filter (EKF) and multi-band difference techniques. The EKF is applied to refine low-precision Position and Orientation System (POS) data under very low image overlap conditions. By establishing error state equations and observation models, the EKF dynamically estimates and compensates for systematic and random errors through iterative recursion, enabling the generation of seamless orthophotos. Subsequently, the multi-band difference method is employed on bi-temporal digital orthophoto maps (DOMs). Differential images for each band are generated, weighted, and fused to enhance change features, thereby facilitating the identification of altered regions in waterway structures. Experimental results from the Jiayu section of the Yangtze River demonstrate that the proposed method effectively produces high-quality image mosaics with only 10% image overlap and accurately detects deformations in regulation structures. This approach offers an efficient and automated solution for waterway inspection and monitoring.