Improved visual tracking using the technique of image pyramid

The advantage of the optical-flow-based visual servo methods is that features of the moving object do not need to be known in advance. Therefore, they can fit for demands of versatile positioning and tracking tasks in the real world. Nevertheless, it is difficult to achieve satisfactory tracking performance for fast moving objects using these standard approaches. The purpose of this paper is to implement a robust image servo scheme that incorporates a sliding-mode control algorithm and the image pyramid technique to track an unknown image pattern in three-dimensional motion. The sliding mode controller deals with possible system uncertainties. The image pyramid builds a group of pictures with different resolutions arranged in layers of a hierarchical structure for the same image. Low resolution images are applied for following a fast moving object to maintain acceptable tracking performance. Switching between layers is performed as the detected motion crosses over a pre-defined maximum critical speed. It is verified that the tracking performance can be greatly improved using the proposed control framework.