The scientific research team of Beijing University of Technology opens up a new field of on-chip optics research

Zhang Jun, an academician team of Beijing University of Technology, pioneered the on chip spectral multiplexing perception architecture, and independently developed the first 100 channel megapixel hyperspectral real-time imaging device in the world, creating the world's highest light energy utilization rate. On November 7, the team's relevant achievements were published in the journal Nature, and the article was entitled "A broadband hyperspectral image sensor with high spatial temporary resolution".

The team proposed the theory and technology of on chip spectral multiplexing perception, which subverted the traditional geometric light splitting, narrowband measurement, and physical output modes, and realized the hyperspectral imaging of on chip broadband alienation control. The team independently developed a hyperspectral intelligent imaging device, which improved the light energy utilization rate from less than 25% to 74.8%. The device weighs only ten grams, and its operating waveband covers the visible and near-infrared ultra wide wavebands (400-1700nm), with an internationally leading space-time spectral resolution of 1024 × 1024@124fps , 96 channels), with completely independent intellectual property rights.

The first author of the paper is Professor Bian Liheng, doctoral student Wang Zhen, and master student Zhang Yuzhe of Beijing University of Technology. The corresponding authors are Academician Zhang Jun and Professor Bian Liheng. Beijing University of Technology is the only completion unit.

Spectrum is called "light gene", which represents the intensity distribution of light signal in different wavebands and represents the intrinsic attribute of the target's reflection/transmission of light. Hyperspectral imaging technology can simultaneously obtain the spatial structure information of the target and spectral information of tens or even hundreds of bands, and can accurately identify the material characteristics of the target, so as to achieve accurate identification of complex environments. It has significant applications in satellite remote sensing, deep space exploration, new quality equipment and many other fields, and is a research hotspot and difficulty pursued by countries all over the world. The existing hyperspectral imaging technology is limited by the traditional mode of geometric splitting and narrow-band measurement. Space, time and spectral resolution are compromised, and the system is large, heavy and difficult to integrate, which seriously restricts its development and application in major fields.

Facing the intelligent and lightweight detection requirements of new quality and new domain applications in the future, Academician Zhang Jun's team innovatively proposed the on chip spectrum broadband sensing architecture. This architecture subverts the traditional discrete geometric light splitting method, and realizes the innovation from complex systems to integrated devices through integrated alienation regulation; Overturn the traditional narrowband measurement mechanism, and realize the leap improvement of optical flux through broadband coupling measurement; Overturn the traditional physical measurement output mode, and achieve high-resolution hyperspectral imaging through intelligent computing.

On chip spectral multiplexing sensing architecture and its working principle

Based on this architecture, the team overcame a series of key technologies such as array based broadband spectral regulation, preparation of hyperspectral intelligent imaging devices, large-scale high-resolution spectral reconstruction, and independently developed the first one hundred channel million pixel hyperspectral real-time imaging device in China, which improved the light energy utilization from typical less than 25% to 74.8%, creating the world's highest record. This device has the advantages of small size (29mm × 29mm × 42mm), light weight (46g), and high intelligence (real-time hyperspectral imaging and accurate target recognition). It can achieve high-resolution spectral imaging in visible near infrared bands. In the range of 400-1000nm, the spectral resolution is 2.65nm, and the spatial and temporal resolution is 2048 × 2048@47fps In the range of 400-1700nm, the spectral resolution is 8.53nm, and the spatial and temporal resolution is 1024 × 1024@124fps The device also has high imaging signal-to-noise ratio (40.2dB), dynamic range (68.71dB) and thermal stability (-60 ℃ -50 ℃).

The device shows broad application prospects in remote sensing detection, life and health, intelligent agriculture, industrial automation and other fields. In the field of remote sensing detection, the team used the device to take high-definition spectral video of the moon surface, to achieve dynamic remote monitoring of observation targets in a weak light environment, demonstrating the excellent light energy utilization and time-space spectral resolution of the device; In the field of life and health, the device realizes dynamic blood oxygen detection and water pollution analysis; In the field of smart agriculture, the device has realized high-precision chlorophyll detection, sugar detection and fruit bruise detection; In the aspect of industrial automation, the device realizes high-precision automatic textile sorting.

This work has opened up a new field of optical research on chip, provided a new method for the development of the next generation of intelligent sensors, promoted the interdisciplinary and integrated development of integrated circuits, electronic information, computers, physics, materials and other disciplines, and helped China's intelligent equipment transform, surpass and become self reliant.

This work has been supported by the National Natural Science Foundation for Distinguished Young Scholars, the National Major Research Instrument Development Project, the Basic Science Center, the General Fund and other projects.

Source: opticsky