When a small object surrounded by a dynamic texture is presented in human peripheral vision, the object is perceived to fade and disappear within a few seconds, under certain conditions. This phenomenon is called perceptual filling-in. Characteristics of filling-in for various kinds of surrounding textures is important to understand the manner of information processing in human vision, because filling-in has been considered that it greatly contributes to capturing external visual information efficiently. From this point of view, we have proposed a model of the filling-in process to address the phenomenon when a small homogeneous area (filling-in target), which is surrounded by spatio-temporal frequency limited random-dot dynamic textures, is presented to an observer's peripheral vision (Proc. IC-EMBS2003). This study reports measurement of time to filling-in for various surrounding dynamic textures that have different spatio-temporal frequency. Applying these results to the proposed model, we estimate the time course of distinguishability of the target from surround (perceptual power). The estimate indicates that if spatial frequency of dynamic textures is low, lower spatio-temporal sensitivity decreases perceptual power more rapidly. However, the opposite property appears if spatial frequency is high.