Motion Tracking With Kalman Filter Prediction and Measurement Update for Robust Position Estimation

Abstract

This paper presents a robust framework for real-time motion tracking using a classical linear Kalman Filter (KF) to predict moving object positions in subsequent video frames. Object tracking constitutes a fundamental task in autonomous systems, robotics, surveillance, and industrial automation applications. Conventional image processing techniques, however, often yield noisy measurements due to motion blur, low illumination conditions, and dynamic background interference. The proposed approach integrates KF into standard image processing to reduce uncertainty in these measurements and improve the accuracy of position prediction. The methodology includes the theoretical fundamentals of Kalman filtering, including state space formulation, measurement, modeling, and recursive predictive update mechanisms. Experimental assessments under various difficult conditions demonstrate a statistically significant reduction in processing error (P <0.05) and an improved robustness compared to the basic line method. Improved arithmetic efficiency and framework accuracy demonstrate broad applicability in visual-based tracking systems in several domains.