A time-analysis of the spatial power spectra indicates the proximity and complexity of the surrounding environment

In this paper, the statistical properties of both simulated and real image sequences, are examined. The image sequences used depict different types of movement, including approaching, receding, translation and rotation. A time analysis was performed to the spatial power spectra obtained for each frame of the image sequences used. Here it is discussed how this information is correlated to the proximity of the objects in the visual scene, as well as with the complexity of the environment. Results show how scene and visual categorization based directly on low-level features, without segmentation or object recognition stages, can benefit object localization and proximity. The work here proposed is even more interesting considering its simplicity, which could be easily applied in a robotic platform responsible for exploratory missions.