期刊名称:Journal of Theoretical and Applied Information Technology
印刷版ISSN:1992-8645
电子版ISSN:1817-3195
出版年度:2021
卷号:99
期号:14
语种:English
出版社:Journal of Theoretical and Applied
摘要:Like any other computer system, robots are susceptible to security problems that could lead to unauthorized disclosure of private information. While it is true that these systems in critical applications (industrial or medical level for example) can result in great damage to equipment and people, in addition to the costs involved with data leakage, domestic applications, in particular service robots, cannot be left aside. A security breach in a system that stores private information puts the security of all individuals in a family group at risk. The communication capabilities of these systems can make them targets for attacks that can seriously compromise the privacy and security of their users. As a sensor system for human environments, it is normal for these robots to incorporate cameras as an important element of their interaction system, similar to the way human eyes do. This research presents an anonymization strategy for all video signals captured by a home service robot as a way to reduce unnecessary personal information captured by these cameras and to increase the confidence of use by users. We propose the use of an algorithm capable of identifying and distorting faces in real-time from the video input of the robot. This process is performed with the original video signal ensuring that no human face is digitized by the robot. The algorithm was evaluated on the ARMOS TurtleBot platform for different operating conditions, demonstrating more than sufficient performance for general robot applications. This strategy will be combined in the future with other systems under investigation by the research group related to source encryption protocols.