An Extended Target Tracking Method Based on Affine Transformation Calibration and Dynamic R‐Table Modelling
Jiayao Wu, Yi Tan, Ce Wang, Xiaohan Mei, Yuanyuan Xie, Shuai Wang, Ping Yang · IET Image Processing · 2026
ABSTRACT When tracking extended targets such as aircraft and drones, the key functional components (e.g., wings, servos, etc.) are often the primary regions of interest. Therefore, continuous and stable tracking of these core components enables targeted behaviour analysis and status assessment. This study addresses the tracking drift problem that occurs when using speeded up robust features (SURF) and generalized hough transform (GHT) methods to track specified locations on extended targets, especially when significant scale and rotation changes occur between consecutive frames. To solve this problem, we propose an extended target tracking method based on affine transformation calibration and dynamic R‐table modelling. First, an R‐table is constructed using the SURF feature points extracted from the previous frame and the corresponding reference point. Then, the affine transformation matrix from the previous frame to the current frame is calculated to calibrate the R‐table. Finally, using the feature points matched between the current and previous frames, the positional relationships indexed in the R‐table are employed to retrieve the coordinates of the reference point. Simultaneously, the R‐table is dynamically updated to achieve high‐precision target tracking. Simulation results show that the proposed algorithm maintains an average centre location error within 1 pixel, even under significant scale and rotation changes between consecutive frames at different resolutions. Moreover, when consecutive frames undergo composite variations (such as scale, rotation, illumination and posture), the average centre location error of the proposed algorithm remains low. Furthermore, the proposed algorithm achieves zero tracking error on both occlusion sequences and fast motion sequences, demonstrating its robustness against a wide range of challenging scenarios.