Speckle-correlation Technique Recovers Images of Obscured Objects in Real Time

Imaging through a light-scattering medium, such as clouds in the sky or tissues in the body, poses special challenges. The scattered light must be reconstructed, typically by using complex optical elements in an environment that is vulnerable to motion and mechanical instability. Computational algorithms are then able to post-process the detected light to generate an image.

A new approach to imaging reconstruction, developed by researchers at King Abdullah University of Science and Technology (KAUST) and the Xiong’an Institute of Innovation, uses speckles to enable clear images of obscured objects, whether static or moving, to be produced in real time.

Previous strategies for reconstructing scattered light have required some knowledge of the object and the ability to control the wavefront of light illuminating it. These strategies have not used directly obtained random speckle patterns for imaging, due to degradation and scattering. Rather, speckles have been seen as noise or chaotic patterns.

Speckle-correlation imaging, which extracts information about the source from fluctuations in the intensity (i.e., speckles) in the transmitted light, could offer an efficient approach to reconstruction. However, many of the technologies based on speckle-correlation require time-consuming computational reconstruction. Also, some information, such as the image orientation and location, is missed.

Illustration of the speckle-correlation imaging system designed by KAUST researchers and collaborators. The technique can produce real-time imaging of static and moving objects using simple, low-cost devices. Courtesy of KAUST/Wenhong Yang.

The KAUST team investigated a way to directly observe self-imaging units of small objects by viewing speckles through diffusers. By carefully examining a speckle, the researchers found that they could obtain clear self-imaging phenomena from a single shot of the speckle image.

By examining the inherent nature of speckles, the researchers were able to develop a self-imaging speckle model and validate it in experiments. This facilitated a visual understanding of speckles and their properties.

The researchers obtained an image directly from a single shot of the speckle image by passing light from a small, standardized test object through a thin diffusing material. By moving the direction of the camera away from the diffuser, the researchers were able to build a 3D image by taking slices through the speckles. When the researchers viewed enlarged sections of these images, they could see reproductions of the test object.

This approach allowed the researchers to see directly through the random diffuser with the naked eye and use real-time video imaging. The orientation of the object could be directly seen and traced in real time.

The new method requires no complex, expensive equipment for the active control of light, and no prior knowledge of the source or diffusion medium. There is no need for iterations or parameter adjustments.

The researchers said that the visibility of directly observed imaging patterns using the new method is equivalent to those processed with direct speckle autocorrelation imaging (SAI). Furthermore, using a simply modified SAI method with efficient joint-filtering, the researchers achieved an imaging quality and resolution comparable to the best results processed by previously reported computational reconstruction methods, according to the team.

“We have developed a strategy of calibration-free, reconstruction-free, real-time imaging of static and moving objects with their actual orientation information,” researcher Wenhong Yang said. “This novel technique only requires simple or low-cost devices, without the post-computational reconstruction.”

The results provide a fresh perspective on diffuser-imaging systems and could inspire new rapid, high-quality imaging applications through scattering diffusers. Optical imaging through scattering light plays a crucial role in many industries, including biomedicine and astronomy.

“Our work presents a significant step in the field of scattering imaging and will shed light on new avenues for imaging through diffusive media,” Yang said.

The research was published in PhotoniX (www.doi.org/10.1186/s43074-022-00080-2).