Snapshot video hyperspectral imaging is a technique that captures both spatial and spectral information of a scene in a single exposure, producing a data cube for each frame in a video sequence. Snapshot methods allow for real-time or near-real-time imaging of dynamic scenes by simultaneously acquiring the entire 3D data cube, in contrast to traditional hyperspectral imaging which scans a scene sequentially. In addition to other sectors, this kind of imaging is widely used in astronomy, machine vision, remote surveillance, medical diagnostics, and agriculture and environmental monitoring.
Snapshot video hyperspectral imaging system featuring the Uniformly Distributed-slit Array coding plate and a SVS-Vistek camera. Image Credit: Image courtesy of University of Chinese Academy of Sciences
Scientists at the University of Chinese Academy of Sciences in Hangzhou, China, recently developed a snapshot video hyperspectral imaging method that increases the speed of full-sample data acquisition, making it superior to other snapshot imaging spectrometers, with additional advantages of being compact, lightweight, and low-cost. Their new system is based on a Uniformly Distributed-slit Array (UDA) coding plate rather than the single slit scanning method found in conventional hyperspectral imaging systems. By using UDA, the scientists found that their approach not only effectively improves the scanning speed of spectrometers but also achieves higher spectral fidelity.
System Design
The optical structure of the University of Chinese Academy of Sciences system calls for the target information being aggregated through a coaxial two-inverted telescope. That data is then encoded by the UDA through a relay optical path with a spectral imager, achieving spectral imaging of the field of view. The telescope can realize the spectral imaging video detection in a spectral range of 450~900 nm. A monochromator is used for spectral calibration, obtaining 60 spectral channels with an average resolution of better than 20 nm and enabling the system to reconstruct the spectrum of interest and acquire a spectral data cube every 0.1 seconds to achieve dynamic detection.
The system is divided into two channels; one channel is a panchromatic imaging system which adopts a SVS-Vistek EXO342-MU3 camera; the other channel is a spectral channel using a detector.
Leveraging the USB3 Vision interface, the SVS-Vistek EXO342-MU3 is a 31.4-megapixel CMOS monochrome camera capable of a maximum resolution of 5464 x 4852 pixels at 12 frames-per-second (fps). In the UDA system, the EXO342-MU3 was set at a lower resolution and functioned at a frame rate of 350 fps.
Testing demonstrated that the proposed method more efficiently collects information and restores the spectral data cube with minimal distortion. Testing objects included the 3D spectral information collection of high-rise building windows and living plants in the laboratory. Results have laid a foundation for the design of a more advanced system with a higher frame rate and resolution.
At present, the system has a signal-to-noise ratio limitation, meaning that it is difficult to apply in scenes with insufficient illumination. Compared to other spectral imaging methods that sacrifice time or computational amounts, however, the proposed method meets the requirements of full sampling, low computational complexity, and hyperspectral imaging speeds.