The development of embedded systems technology has changed the history of automotive world, bringing new devices including modern access vehicle's control. In these systems' recent history, industry has introduced the concept of Passive Access or Passive Entry (PASE) system, allowing the user to control vehicle's doors opening and closing without the need for pressing any button. These systems are based on radio frequency devices with exchange of encrypted information with the Remote control Key (RKE), which assures user's authentication. In spite of the comfort provided by this technology, there is the possibility of attacks against PASE and RKE authentication, exploring access security flaws, thus requiring constant research and development and improvement on these devices. This work proposes and evaluate Face Recognition (FR) for user authentication integrated with PASE under unconstrained environments and illumination variation as alternative to RKE based systems. A Design Science Research (DSR) based methodology was used to support the instantiation of a FR framework, which was validated using Receiver Operating Characteristic (ROC) curves.A vehicle prototype already mounted with PASE system was integrated with FR algorithms instantiations as artifacts: Eingenfaces, Fisherfaces, 2DFLD (2D-LDA) and VIOLA-JONES detector, supported by Near Infrared Cameras (NIR).The results were evaluated regarding computational cost (memory and processing time) of selected Face Recognition algorithms authentication and compared with available integrative capacity of automotive embedded devices.In summary, from the experiments and instantiations supported by DSR method and also confirmed during all test-cases executed, this work concluded it was feasible to integrate FR algorithms and Passive entry systems, confirming also VIOLA-JONES detector in conjunction with Infrared LEDs to overcome illumination variation under unconstrained environments.Among FR algorithms, Fisherfaces has been confirmed as the best option due to its stability, low memory consumption, less training samples and adequate overall execution speed which is compatible with embedded micro-controllers.
展开▼