On-line object tracking is an essential technology in computer vision. Object tracking systems need to reduce their energy consumption because the technology is increasingly being utilized for battery-operated systems, e.g., driving assist systems, smartphones, drones and so on. To tackle this problem, dynamic frame-rate optimization has been proposed. This approach optimizes the frame-rate on the basis of target object speed by taking into account the energy trade-off between the image capturing and tracking processes. In order to improve tracking accuracy, the approach selects a frame-rate based on a specific fixed value. However, the required parameters are different depending on the scene and content of the input video. In this paper, we propose a method to adaptively select parameters. Simulation results show the energy consumption is reduced by up to about 65.0%, and 45.0 % on average without critical tracking accuracy degradation.