Target tracking in nonuniform illumination conditions using locally adaptive correlation filters

An accurate method for tracking the position and orientation of a moving target in nonuniformly illuminated and noisy scenes is proposed. The approach employs a filter bank of space-variant correlation filters which adapt their parameters accordingly with local statistics of the observed scene in each frame. When a scene frame is captured, a fragment of interest is extracted from the frame around predicted coordinates of the target location. The fragment is firstly preprocessed to correct the illumination. Afterwards, the location and orientation of the target are estimated from the corrected fragment with the help of the filter bank. The performance of the proposed system in terms of tracking accuracy is tested in nonuniformly illuminated and noisy scene sequences. The obtained results are discussed and compared with those of similar state-of-the-art techniques for target tracking in terms of objective metrics.