Advances in electronics, sensor technologies, embedded hardware and software are boosting the\udapplication scenarios of wireless sensor networks. Specifically, the incorporation of visual capabilities into\udthe nodes means a milestone, and a challenge, in terms of the amount of information sensed and\udprocessed by these networks. The scarcity of resources – power, processing and memory – imposes strong\udrestrictions on the vision hardware and algorithms suitable for implementation at the nodes. Both,\udhardware and algorithms must be adapted to the particular characteristics of the targeted application. This\udpermits to achieve the required performance at lower energy and computational cost. We have followed\udthis approach when addressing the detection of forest fires by means of wireless visual sensor networks.\udFrom the development of a smoke detection algorithm down to the design of a low‐power smart imager,\udevery step along the way has been influenced by the objective of reducing power consumption and\udcomputational resources as much as possible. Of course, reliability and robustness against false alarms\udhave also been crucial requirements demanded by this specific application. All in all, we summarize in this\udpaper our experience in this topic. In addition to a prototype vision system based on a full‐custom smart\udimager, we also report results from a vision system based on ultra‐low‐power low‐cost commercial imagers\udwith a resolution of 30×30 pixels. Even for this small number of pixels, we have been able to detect smoke\udat around 100 meters away without false alarms. For such tiny images, smoke is simply a moving grey stain\udwithin a blurry scene, but it features a particular spatio‐temporal dynamics. As described in the manuscript,\udthe key point to succeed with so low resolution thus falls on the adequate encoding of that dynamics at\udalgorithm level
展开▼
