Visual object tracking has been challenged in computer vision because objects often undergo significant representation changes caused by occlusion, illumination variation, scale variation and rotation. Unfortunately, recent trackers do not focus on dynamic characteristics of tracking object. In this paper, we present a strategy based on visual attention feature (VAF), which uses features related to visual attention mechanism and object movement distribution. Firstly, Visual attention feature has been constructed and extracted. Then, concepts of visual attention feature have been illustrated. Next, it is used to improve tracking performance based on classic discriminative method and parallel computing platform. Finally, statistical results demonstrate effectiveness and accuracy of our object tracking strategy based on VAF. In addition, experimental results on OTB-2013 and OTB-2015 benchmark datasets show that our strategy with parallel candidate region generating algorithm improves robustness of recent object tracking methods against many challenging attributes and suffices to reach state-of-the-art performance. © 2018 Elsevier Inc.