Summary:
UCLA researchers in the Department of Electrical and Computer Engineering have developed a novel platform technology to facilitate the design of all-optical visual processors, which can be used to perform advanced computational tasks at the speed of light.
Background:
Information processing via light is a cutting-edge field among optics and photonics researchers. It holds the potential for rapid, energy efficient computation that is vital for future computing and for emerging technologies such as autonomous vehicles, where ultrafast processing of natural signals is of utmost importance. However, optical processors have not replaced traditional electronic computing for most applications. Since natural lighting is primarily composed of spatially incoherent light, processing visual information under these conditions is essential for imaging and sensing applications. There is an urgent need to develop advanced optical processing systems that can interpret information derived from natural lighting conditions rapidly. By creating better hardware and software solutions, optical processors could be more broadly adopted by many industries.
Innovation:
Professor Ozcan and colleagues have developed a method for designing all-optical universal linear processors of spatially incoherent light. Their optical processors are comprised of a set of structurally engineered surfaces. These surfaces exploit the successive diffraction of light to perform linear transformations of the light field without using external digital computational power. They further show that using deep learning-based methods they can perform any arbitrary linear transformation using just the optical intensity of the light source. In addition, these optical processors can be used to perform transformations in parallel using broadband light. This potentially reduces the hardware complexity required for an optical processing system. The ability to process incoherent information optically, without reliance on electronic or digital systems, marks a significant advancement in the field of optical computing.
Potential Applications:
• Optical processors
• Visual computing systems
• Computational microscopy
• Imaging platforms
• Autonomous vehicle navigation
Advantages:
• Processing of spatially-incoherent light
• Works with phase and amplitude optical features
• High accuracy
• Compatible with any linear intensity transformation
• Can be extended to complex-valued transformations
Development-To-Date:
Researchers have evaluated the performance of this platform by using it to successfully classify handwritten digits under spatially incoherent illumination with an accuracy of over 95%.
Related Papers:
• Rahman, M., Yang, X., Li, J., Bai B., and Ozcan, A. Universal Linear Intensity Transformations Using Spatially-Incoherent Diffractive Procesors. Light: Science & Applications 2023 https://www.nature.com/articles/s41377-023-01234-y
Press Release:
Universal linear processing of spatially incoherent light through diffractive optical networks
Reference:
UCLA Case No. 2023-192
Lead Inventor:
Professor Aydogan Ozcan, UCLA Department of Electrical and Computer Engineering