Eye-inspired imaging chip will extend video capabilities

Researchers are developing a dynamic imaging chip inspired by the human retina that will be capable of capturing high-performance video at low bandwidth and power consumption.

The ‘silicon retina’ is being developed by a European consortium and will have immediate applications for the electronics industry and for machine vision — with artificial prosthesis for blind people representing a long-term goal.

Charge-coupled devices (CCDs) and complementary metal-oxide-semiconductors (CMOSs) are the most commonly used imaging sensors and have found their way into a variety of devices including mobile phones and laptops.

‘If you’re taking photos that’s fine, because you need every single pixel,’ said project partner Dr Konstantin Nikolic of Imperial College London.

‘But if you’re making a continuous film, what happens is there is massive redundancy of information. If you’re filming at 40fps, each frame consists of millions of pixels, yet there is very little different from frame to frame.’

Nikolic explained that the retina acts as a filter by only sending on salient spatial and temporal information to the brain via the optic nerve.

‘It’s like a microprocessor that compresses data — it’s very efficient at reducing the bandwidth to transmit the information. Say we have an input of about 100 million pixels on the retina, the output is only about a million pixels.’

By understanding how the retina achieves this, the group will be able to make imaging chips that have high performance with low power consumption. They will also be particularly useful for simple tasks such as autonomous video tracking, which could have applications in monitoring elderly people for falls or traffic surveillance.

Nikolic is, however, cautious about the potential for vision augmentation. In most blind people, the photoreceptive cells — the cones and rods — are damaged, but the cells that convert the light into an electrical signal and then send it to the brain — the ganglia — are intact. Thus some research groups have tried to effectively ‘bolt on’ an imaging sensor to the ganglia, with varying degrees of success.

‘This stimulation is interpreted by the brain as noise, so all people can see is white dots called phosphenes. There’s is no proper visual perception as we know it, it’s very crude. Ganglia cells are very versatile, the process is unknown, it’s a huge research problem.’

In addition to Imperial College, the ‘SeeBetter’ consortium involves researchers at IMEC in Belgium, the Friedrich Miescher Institute for Biomedical Research and the University of Zurich, both in Switzerland.