Penn State researchers have created a new device that produces images by mimicking the red, green, and blue photoreceptors and neural network found in the human eye. According to Kai Wang, assistant research professor in Penn State's Department of Materials Science and Engineering, “we borrowed a design from nature—our retinas contain cone cells sensitive to red, green, and blue light and a neural network that starts processing what we see even before the information is transmitted to our brains. ”
“The colorful world we can see is a result of this natural process.”
To achieve this in an artificial device, the researchers developed a new array of sensors made up of narrowband perovskite photodetectors similar to our cone cells. They combined it with a neuromorphic algorithm similar to our neural network to process the data and generate highly accurate images.
Photodetectors are crucial to cameras and many other optical devices as they convert light energy into electrical pulses. According to the researchers, the reds, greens, and blues that make up visible light are just a few examples of regions of the light spectrum that can be focused using narrowband photodetectors.
According to Wang, in this work we developed a revolutionary method for creating perovskite material that is sensitive to only one wavelength of light. “We developed three different perovskite materials that are sensitive only to red, green, or blue wavelengths,” the author says.
According to experts, this technology could be a way to avoid the use of filters in modern cameras that reduce resolution, increase costs and complicate production processes.
The cameras use silicon photodetectors that absorb light but cannot distinguish between hues. Red, green, and blue light are separated by an external filter that only allows one color to enter each area of the light sensor and wastes two-thirds of the incident light.
“Some information is lost when the light is filtered out, but with our design this can be avoided. Therefore, we think that this work may represent a potential camera detection method that can help achieve a higher spatial resolution.
According to the researchers, the new devices generate power while absorbing light due to the use of perovskite materials, potentially paving the way for battery-free camera technology.
According to Luyao Zheng, a postdoctoral researcher at Penn State, “the device structure is similar to solar cells that use light to generate electricity.” “When you shine a light on it, it will produce a current. We don't need to use energy to gather this information from light, just like our eyes do.
As a result of this research, further progress can be made in the biotechnology of artificial retinas. According to the researchers, devices based on this technique could one day be used to replace damaged or dead cells in our eyes in order to restore vision.
The researchers write in the journal Science Advances that the discoveries reflect multiple important advances in the development of perovskite narrowband photodetection devices, from materials synthesis to device design and systems innovation.
Perovskites are semiconductors that produce electron-hole pairs when light strikes them. Electric currents are produced by sending these electrons and holes in opposite directions.
In the thin-film perovskites of this study, holes flow through the material more rapidly than electrons, which have significant instability in electron-hole transport. By adjusting the architecture, or how the layers are stacked, in unstable perovskites, the researchers discovered they could exploit properties that would enable the materials to function as narrowband photodetectors.
These materials were used to make a sensor array and a projector was used to illuminate an image through the apparatus. For signal processing and image reconstruction, data from the red, green, and blue layers was entered into a three-sub-layer neuromorphic algorithm. One type of computing technology known as neuromorphic algorithms aims to mimic how the human brain works.
Wang added that they have tried various data processing methods. “We tried combining the three color layer signals directly, but the result was not very clear. However, when we apply this neuromorphic processing, the image looks much more like the original.
Stating that the program is similar to the neural network found in the human retina, the researchers said their findings may therefore shed new light on the importance of these neural networks for vision.
According to Wang, by combining our device and this method, we can show that neural network capability is crucial in visual processing in the human eye.
Günceleme: 13/05/2023 21:58