BEIJING, June 1, 2020 — To reduce noise in a thin, lensless camera, researchers from Tsinghua University and MIT have built an imaging system with a Fresnel Zone Aperture (FZA). In addition to introducing an optical Fresnel element into the camera, the researchers used a compressive detection algorithm to improve the quality of the reconstructed images. The FZA was placed 3 mm in front of a CMOS image sensor. The signal recorded by the CMOS sensor was reconstructed by the compressive detection algorithm with total variation denoising to generate an improved image of the object.
Although the fresnel zone plate can be used as a lens-like imaging element, its long focal length cannot support a thin structure. “If we can get the light field on the output plane of [the] objective, the light field on the focal plane could be reconstructed by digital propagation,” said researcher Jiachen Wu. “Then the length of [the] imaging system can be significantly shortened.
It was a challenge to get the light field because the sensor could only register intensity and not phase information. Since holography can record an interference pattern to reproduce a light field, the researchers decided to extend the concept of holography for incoherent illumination to their lensless camera.
At MIT and Tsinghua University, researchers have developed a new lensless camera using a Fresnel zone aperture. Courtesy of J. Wu, H. Zhang, W. Zhang, G. Jin, L. Cao and G. Barbastathis.
Using the FZA imaging system and CMOS sensor setup, the researchers took a series of photos that were displayed on an LCD screen. Incident light rays from a point on the object passed through the FZA and cast the shadow of the FZA on the sensor. The resulting image from the camera sensor showed a distinct fringing feature similar to that found in a hologram.
When they discovered that the shadow of the Fresnel zone plate had the same shape as a point source hologram, the researchers assumed that the object could be encoded in a hologram using a Fresnel zone plate. under the inconsistent lighting. The image could then be reconstructed by backpropagation.
The quality of the reconstructed images is greatly improved by the use of a compressive detection algorithm. Courtesy of J. Wu, H. Zhang, W. Zhang, G. Jin, L. Cao and G. Barbastathis.
To reconstruct an image from a single “hologram” and avoid a twin image effect, the researchers introduced a total variation constraint in the image reconstruction. In this way, the twin image, which did not respect the constraints of total variation, was eliminated. Thanks to the compression detection algorithm, single-shot imaging can be obtained without any calibration.
The iteration process of the compressive detection algorithm. Courtesy of J. Wu, H. Zhang, W. Zhang, G. Jin, L. Cao and G. Barbastathis.
This computer imaging architecture could improve the quality and reduce the cost of lensless cameras. The FZA pattern could be placed on the glass covering the sensor, integrating the camera and sensor and simplifying manufacturing. This thin, lensless camera could be used in ultra-thin smartphones, home security cameras, and self-driving vehicles.
“We are trying to open a door for high-quality, noise-free lensless cameras,” said Professor George Barbastathis. “The presented technique provides a prototype for the integration of cameras and smart devices. Through a partnership between academia and industry, this technique could become practical.
The research has been published in Light: science and applications (www.doi.org/10.1038/s41377-020-0289-9).