Evaluating low-cost ARM SBCs as video-capture nodes in a vehicular panoramic video recording system
Chua, Koon Ming
Date of Issue2014
School of Computer Engineering
An increasing numbers of vehicles are being fitted with cameras to provide a video log that may be used in case of accidents on the road. These cameras typically capture only one view (forward-looking) with a limited field of view and poor low-light performance. Furthermore, they typically do not have much redundancy. Solutions do exists in the market that solves such problems, such as those from PointGrey. However, devices like these are costly. But with the emergence of low-cost SBC (Single Board Computers) gaining popularity, and software becoming more prominent, it is becoming feasible to have a 360 FOV in car DVR. Therefore, in this project, capabilities of low-cost ARM-based single board computer are investigated for their suitability as the primary processor for video-capture nodes that will be networked into a panoramic DVR system. This includes (but is not restricted to) the optimizing boot time (<10s startup), driver availability for hardware-accelerated video capture and encoding. Two different boards, the Raspberry Pi and Odroid X was compared in this study using Arch Linux as the base operating system and GStreamer to stream video feeds to another PC. Performance of H.264 and MJPEG over RTSP/UDP were evaluated and there were mixed findings. Mixed findings in a sense that the choice of an SBC would still ultimately have to depend on usage scenarios. However, in general, when going for cost restricted scenarios the Odroid would be the choice, whereas when going for quality restricted or networked scenarios, the Raspberry Pi and its native camera would be the choice.
DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
Final Year Project (FYP)
Nanyang Technological University