We address the problem of continuous skyline computation on highly dynamic moving objects (i.e. objects with dynamic dimensions move in an unrestricted and unpredictable fashion), which is quite a different scenario from existing literatures on skyline algorithms. We propose a continuous skyline query processing strategy for static query point, and the main idea is as follows: (1) The work space is divided into lots of regular grids, and the valid objects are indexed by this data structure. (2) Some grids are organized as the influence region, while the rest compose of the free region. The algorithm achieves low running time by handling movements only from objects that fall in the influence region, while data changes in the free region are omitted with correctness guarantee. (3) The initialization module adopts an efficient method to obtain the initial result without having to process all the data points; after that the maintenance module updates the change of skyline and influence region dynamically when data changes. We analyze the space and time costs of the proposed method and conduct an extensive experiment, which indicates that our gridbased algorithm is efficient and significantly outperforms existing methods adopted for the application.