Insect eyes see the dark half

By mimicking how insects see, Australian research could lead to digital videos in which every detail can be seen, even in conditions with a huge contrast in brightness. The technique would solve a critical problem for surveillance cameras, where the clarity of images is everything.

According to Dr Russell Brinkworth, a postdoctoral research fellow at the University of Adelaide, even a tiny insect brain can outperform any current artificial visual system. Like humans, they can see detail in light and dark at the same time.

Traditional cameras use a single average light setting to control the brightness of an image, which works if there are similar levels of lighting over the whole scene. But it does not work so well if some parts are much brighter than others.

In nature, the individual cells of the eye adjust to a part of the image independently in order to capture the maximum amount of information about the scene. This means that even in difficult lighting conditions, such as a person standing in front of a window, you can see both the person’s face and the scenery outside at the same time, something a traditional camera cannot do.

By recording from cells in the brains of insects Brinkworth and his colleagues have shown it is possible to determine exactly how animal eyes work, and to reproduce the process using computer software and hardware.

It is a fundamentally new way of thinking about vision technology, Brinkworth says. Eyes are designed to detect movement and they cannot actually make out anything that is stationary relative to them. The image humans see is created by the movement of their bodies and the flickering of their eyes.

Brinkworth hopes his research will help cameras resolve detail in light and dark and detect moving objects. It could also rapidly compress and transmit video at incredible speed, and detect and measure the speed of very small objects moving in the distance.

Brinkworth’s software can already enhance existing video footage. The next step is to wire this into existing camera sensor technology. The software would be written to a computer chip that would sit between the sensor and the digital converter.