摘要：Owing to recent developments in surveillance-system technology, there is growing interest in the construction of intelligent surveillance systems aimed at public facility management. Intelligent surveillance systems detect abnormal phenomena through monitoring by managers, and require further intelligent analysis of pictures. Such systems may have difficulty analyzing pictures that contain certain barrier elements. Therefore, preprocessing technology is necessary to improve these barrier elements before analysis of the pictures. Barrier elements in surveillance systems aimed at facility management include the unexpected appearance of obstacles or areas of shadow within the facility. Existing research on shadow areas does not make good use of any prior information that the intelligent surveillance system might have. It is mainly concerned with aerial photograph and therefore has limited application to facility management. Accordingly, this study proposes a technique that is suitable for application to facility-surveillance systems to predict and improve shadow areas in picture data. Further, this study examines the effectiveness of the proposed technique by means of an experiment.
