2024 Volume 80 Issue 20 Article ID: 24-20108
Recently, with the commercialization of augmented reality devices and the launch of the 5th generation mobile communication network (5G), the applicability of augmented reality in various spaces has been increasing. In particular, for road use, it is necessary to have a view management that can display a variety of data without disturbing the view of the surrounding space. In this paper, we propose a calculation method of important regions to be viewed based on temporal and spatial high-frequency components and object detection results by Convolutional Neural Network (CNN) in first-person view video. In order to verify the effectiveness of the proposed method, we conducted an experiment using actual first-person view video images. As a result of the experiment, we confirmed the effectiveness of the proposed method by calculating various regions to be seen with general purpose and high accuracy.
