Motion-enhanced sensor captures ultra-high-resolution images, overcoming a pixel miniaturization bottleneck

 

Digital image sensors (DIS), devices that capture images by converting light patterns into electrical signals, are integrated in many contemporary electronic devices, including smartphones, digital cameras and some medical instruments. These sensors rely on tiny light-sensitive units called pixels, which record brightness and color.

To increase the resolution of the images captured by DIS, engineers have been increasing the density of pixels, packing more of these tiny units into smaller areas. Since smaller pixels collect less light, this approach can weaken useful signals and increase random fluctuations (i.e., noise). This ultimately results in grainy and less detailed images, particularly in low lighting conditions.

Researchers at Tsinghua University have introduced a new approach to increase the resolution of images captured by a DIS. Their proposed method, outlined in a paper published in Nature Electronics, entails integrating an image sensor with a small mechanical device that slightly shifts the sensor's position while it captures images, allowing it to gather more visual information.

"The inspiration for this work stems from a critical bottleneck currently facing DIS," Xiaoguang Zhao, senior author of the paper, told Tech Xplore.

"Although pixel scaling continues, pixel miniaturization is approaching its physical limits due to signal-to-noise ratio constraints. However, emerging applications in multidimensional sensing, such as light-field and spectral imaging, are driving an exponential demand for a higher Spatial Bandwidth Product (SBP), defined as the product of spatial resolution and field of view."

A moving sensor that captures high-resolution images

As part of their study, Zhao and his colleagues set out to develop a chip-scale image sensor that would break the limits of previous DIS devices, reaching higher resolutions. To do this, they first devised a new universal imaging paradigm that introduces a way to enhance the spatial bandwidth product (SBP) of sensors.

The SBP is a property of image sensors that indicates the richness of visual details that they can capture overall. To improve this property, the team proposed combining pixel miniaturization with the integration of a microelectromechanical system (MEMS).

"Our developed MEMS-based SBP-enhanced image sensor (MSEIS) integrates a DIS chip onto a MEMS actuator via flip-chip bonding," explained Zhao.

"By precisely controlling the MEMS actuator, we achieve sub-pixel spatial position modulation of the DIS on the focal plane. This effectively decouples the 'sampling period' from the 'pixel size.' The sampling period is no longer limited by the pixel pitch but is determined by the displacement accuracy of the actuator, which typically reaches the nanometer scale for MEMS actuators."

In contrast with other approaches to improve the resolution of cameras, the team's solution can also be applied on a chip-scale, enabling the development of smaller sensors that can be easily integrated with other electronic components. The researchers used their method to develop a new image sensor and tested it in a series of experiments, which yielded very promising results.

"Experiments demonstrate that our approach can enhance the SBP by approximately 33.7 times and can be seamlessly integrated into existing imaging systems," said Zhao. "We utilize solder bumps for electrical interconnection, and flip-chip bonding the DIS onto the movable MEMS platform, which may serve as a new scheme for the heterogeneous integration."

A new route to advance imaging technologies

The team's new paradigm for improving the resolution of DIS devices could open new possibilities for the development of advanced imaging technologies. In the future, it could be used to further improve the resolution and clarity of images captured by image sensors.

"We established an extended imaging theoretical model based on Fourier optics for the spatial modulated imaging paradigm, quantitatively describing the SBP enhancement effect of spatial modulation imaging," said Zhao. "Furthermore, we constructed a chip-scale spatial modulated image sensor, successfully achieving an SBP enhancement of approximately 33.7 times and reducing the equivalent sampling period from 3.6 μm to 0.62 μm."

The new sensor developed by Zhao and his colleagues could eventually be used for various advanced imaging applications, including point target localization. This entails the precise tracking of stars or of other objects that are difficult to capture in images.

"As part of our future studies, we plan to introduce optimized scanning strategies (e.g., Lissajous scanning) to reduce response time and adapt to higher-speed imaging scenarios," added Zhao.

"We are also exploring non-uniform sampling strategies, such as Monte Carlo sampling, to increase the flexibility of SBP enhancement. Finally, we wish to develop wafer-level MEMS-CMOS co-fabrication technology to enable the mass production of highly integrated and reliable MSEIS chips."