Pavel Cheremkhin, PhD (Physics and Mathematics), an alumnus of the MEPhI National Research Nuclear University, knows everything about digital holograms – or almost everything. A team from LaPlas MEPhI Laboratory of Photonics and Optical Information Processing has recently published a study, Digital Hologram Reconstruction via Multibranch Neural Network, in Applied Sciences. The contributors – Associate Professor Pavel Cheremkhin, postgraduate student Andrei Svistunov, doctoral student Dmitry Rymov and Professor Rostislav Starikov – proposed a new perspective on, as well as new applications for holograms.
The theory of holography was developed in 1947 by Hungarian-British scientist Dennis Gabor, but actual holograms could only be made with the creation of the laser. A beam of coherent laser light is directed onto an object, with part of it split off as a reference beam. The two beams interfere with each other; the interference pattern is recorded on a plate, which is called a hologram.
In 1962, Soviet optical physicist Yury Denisyuk produced the world’s first three-dimensional hologram. Later, this kind of white-light reflection holograms became widespread and even penetrated our daily lives. For example, an Australian company recently opened a unique Hologram Zoo in Brisbane, where all animals, including long-extinct ones, are holograms. Of course, applications for holography go far beyond entertainment; holograms help make our lives better and more comfortable, and solve scientific, medical, and environmental problems. At first, holograms were recorded on photographic plates, but with the development of digital technology, it became possible to save them as computer files. The tech sounds simple when explained like this. In reality though, creating a high-quality hologram is a challenging and time-consuming process.
A hologram is an optical image of an object that contains information not only about its shape, but also about its structural features. To reconstruct a 3D scene, they use computer simulation or optical devices such as spatial light modulators. Importantly, holograms reconstruct objects with high precision despite any negative factors present during recording. This is what Pavel Cheremkhin focuses on in his research.
“Digital holography is a technology for recording and processing data on the object’s structure and characteristics in a two-dimensional or three-dimensional form. The technology was facilitated by the development of optical-digital methods for recording images and their computer processing. Photographing the resulting image enables us to restore and supplement information about the object, as well as explore its important features. For example, by using this method to examine a material, we can detect inclusions, evaluate its density or spot the slightest roughness on its surface. Thanks to high-speed recording of digital holograms, we can easily track the dynamics of changes in an object,” Pavel Cheremkhin explains.
The technology can be used in 3D monitoring systems, in nano- and microelectronics, and in the study of macro- and micro-objects, such as microplastics. For example, it can be useful in exploring the Great Pacific Garbage Patch. Also known as the Pacific trash vortex, this patch of plastic waste is expanding rapidly, posing a massive threat to the ocean ecosystem. Digital holography can help determine the exact concentration of plastic in each specific part of the patch, measure its density and particle sizes, and track any changes. This should contribute to the development of the right strategy for cleaning the ocean.
All it takes is a laser and digital cameras that will register the light reflected from the water surface and microplastics particles. The resulting hologram will show how plastic interacts with the environment, allowing scientists to analyze all reactions in real time. In challenging cases, such as the microplastics patch in the ocean, the image of an object may be distorted by noise (accidental changes in brightness or color data), which affects a hologram’s quality and makes it unclear or indistinct on the general background.
MEPhI scientists set a goal to recreate quality images of digital holograms for difficult cases, such as when multiple objects are spread across the entire volume and the noise level is high. As an additional advantage, the method proposed by the scientists relies on neural networks to expedite calculations and reduce the noise.
“It is possible to eliminate the noise by registering an entire set of holograms; however, this strategy affects the quality of research. We proposed a new method of recreating hologram images by using neural networks and trained our neural network to recreate a multitude of two-dimensional cross-sections of a 3D scene. It takes centiseconds and even milliseconds to calculate the recreation. The resulting images are clear of noise and provide a rather accurate position of each micro-object. We can measure their quantity, size and density. When registering a holographic video, we can watch objects change and track their dynamic in real time,” Pavel Cheremkhin explains.
Digital holograms have plenty of applications and, the scientist believes, they will acquire more common uses in the not so distant future. In healthcare, holograms can be used during surgeries and clinical analysis (for example, to count red blood cells, measure their size, spot any deformed cells – and all this with high precision). The method can be helpful when analyzing dynamic and fast-moving processes that require promptly locating particles in a dispersed medium. Digital holography can make textbooks and visual aids more attractive and informative or increase the precision of architecture designs and device assembling. And certainly, who wouldn’t enjoy communicating with friends via holographic video link-ups and see the other person’s holographic image?
More information about recreating 3D images using digital holography and the HoloForkNet method can be found in the research paper, Digital Hologram Reconstruction via Multibranch Neural Network. The research project received a grant from the Russian Science Foundation. A similar research project is conducted under the Priority 2030 program.