Artificial Eye

Researchers are looking to the human eye to address some of the drawbacks of existing imaging systems. By mimicking the way in which mammals process visual information, they have developed a new class of devices, known as neuromorphic sensors, that can dramatically reduce the power demands and processing needs of cameras for surveillance and other uses.

Conventional cameras capture video by storing a series of frames. Whether the scene is relatively static or dynamically changing, the same amount of data is collected. Neuromorphic sensors, on the other hand, sample different parts of the scene at different rates depending on changes in light conditions and are not restricted to a fixed frame rate. This frees up processing power and allows the fastest-moving sections to be captured in greater detail.

The reduced energy required to run such sensors will make them useful for applications where it’s impractical to recharge a battery, including surveillance drones and retinal implants. Future research will focus on how the dynamic visual sensors could be used to share high-quality images between machines or upload them to the cloud.

For information: Maria Martini, Kingston University, Department of Science, Engineering & Computing, Penhryn Road, Kingston upon Thames, Surrey KT1 2EE, United Kingdom; phone: +44-020-8417-9000; email: m.martini@kingston.ac.uk; website: http://www.kingston.ac.uk/