To show a multi-directional scene situation with switching different views, we have to keep a screen layout of a remarkable object for intelligible image media. In this paper, we propose a method to navigate a viewer to one of objects in the scene by keeping its screen layout for switching multiple views. In order to keep the layout, we developed "virtual swiveling" by using fixed-viewpoint pan-tilt-zoom camera, based on area and center of gravity of extracted objects from captured images. We also show several experimental results to confirm an effectiveness of our method.