Tracking of Object with Occlusion based on Normalized Cross Correlation and Kalman Filter Estimation

This paper proposes an algorithm that uses Normalized Cross Correlation (NCC) and Kalman Filter (KF) for object tracking. Occlusion during visual tracking reduces the tracking performance and may even lead to track loss. Partial occlusion may cause target template get corrupted. During complete occlusion, false target tracking may take place as actual target is not present in the scene. This paper proposes a method combination of correlation tracking based on NCC and Kalman filter based Outlier Detection algorithm for robust tracking during occlusion. The outlier detection algorithm works based on adaptive bound for innovation in Kalman filter and detects presence of occlusion. Once the outlier is detected, the prediction of Kalman filter estimation is used to predict the target position. The results of track during occlusion in presence of false target show the robustness of this approach.