Object Tracking in SWIR Imaging Based on Both Correlation and Robust Kalman Filters

Short-wave infrared (SWIR) imaging has significant advantages in challenging propagation conditions where the effectiveness of visible-light and thermal is limited. Object tracking SWIR particularly difficult due to lack color information, but also because occlusions maneuvers tracked object. This paper proposes a new algorithm for object imaging, using kernelized correlation filter (KCF) as basic tracker. To overcome occlusions, use Kalman predictor method expand search area. Expanding area helps better re-detection after occlusion, leads occasional appearance errors measurement data that can lead loss. These be treated outliers. cope with outliers, Huber’s M-robust approach applied, so this robustification by introducing nonlinear influence function estimation step. However, robustness outliers comes at cost reduced estimator efficiency. make balance between desired efficiency resistance adaptive M-robustified proposed. achieved adjusting saturation threshold detection confidence information from KCF Experimental results on created dataset video sequences indicate proposed achieves performance than state-of-the-art trackers maneuvering presence occlusions.