VIS / SWIR HyperSpectral Imaging

The above video shows the scanning/acquisition of images through a selected range of VIS + NIR wavelengths. Once a dataset is acquired, it can be used to create a photorealistic RGB image (as shown in the above gallery). Spectral information from the ENVI-compatible datacube is readily available for analysis, classification and/or segmentation.

Hyperspectral Imaging technology enables the acquisition and analysis of images that reveal the spectral distribution of light that is collected at each pixel site within an image. In H.S.I., both spectral and spatial information about a sample are obtained. It is analogous to having spectrometric data for each pixel in an image.

A hyperspectral image is a set of image layers, with each layer representing the spatial intensity distribution at a particular wavelength. This data can be represented as a λ-axis that is orthogonal to the conventional (x,y) axes used to spatially represent image data. The combination of all the acquired spectral layers is called a hyperspectral data cube.

Several different methods can be used to produce hyperspectral image datacubes, each representing a particular tradeoff. Each applies different techniques to spectrally filter the light and capture the spatial and spectral image data. Each of the different methods of acquiring hyperspectral image datacubes emphasizes one or more of the following key differentiators:

acquisition speed – the time required to capture the hyperspectral data cube
snapshot capability – the ability to acquire a hyperspectral image, without scanning
spatial resolution – the size of the pixel array, similar to regular photography
spectral resolution – the number of frequency bands (or layers) each hyperspectral image contains.

The cameras described on this page employ imec's proprietary technology in which 10~15nm FWHM Fabry-Perot band-pass spectral filters are deposited on each pixel of a VNIR (Visible + NIR) CMOS -or- SWIR InGaAs imager. Two different types of cameras, both based on imec's Hyperspectral Imaging technology, are described on this page:

imec Snapscan technology - ideal for research environments, this is a spectral-scanning method which achieves the highest spectral and spatial resolution. The tradeoff is that it can take a few seconds to scan through a range of wavelengths.

imec Snapshot technology - ideal for deployment in high-throughput production environments as well as for vehicle-mounted applications (including UAVs), this method prioritizes acquisition speed. The tradeoff for achieving spectral data at real-time video rates is that it collects data at selected wavelengths and that its spatial resolution is lower than that of a scanning system.

For more information about the above technology options please click on the tabs below. To review the different models of Snapscan and Snapshot that are available, please scroll further down on this page.

Snapscan: introduction Snapscan: Theory of Operation Snapshot: introduction Snapshot: Theory of Operation Microscopy Example imec Video

Snapscan: introduction

imec Snapscan technology - ideal for research environments
In research applications, it may be critical to capture hyperspectral datacubes over a wide range of wavelengths. Imec's Snapscan technology permits the acquisition of hyperspectral imaging datacubes with high spatial resolution AND high spectral resolution. The tradeoff is that this requires a scanning process that takes several seconds to acquire a hyperspectral imaging datacube. Often this allows scientists to identify the specific wavelengths of interest from images taken over many samples. This information can subsequently be used to develop a real-time Snapshot imaging system that is limited to the specific wavelengths of interest.

Snapscan: Theory of Operation

Snapscan cameras are used by researchers to spectrally inspect stationary samples by generating HyperSpectral image datacubes with high spectral resolution AND high spatial resolution. In Snapscan cameras, Fabry-Perot band-pass filters are deposited on the image sensors in a striped pattern of filter bands, spanning the broad spectral range of the underlying VNIR or SWIR imager.

The cameras are c-mount compatible, which allows their use in microscopy or with conventional camera lenses. Snapscan cameras generate HSI datacubes in seconds by translating the image sensor with its striped pattern of filter bands through the image circle that is projected by the lens or microscope optics onto the focal plane. Pixel values are read out and can now be rearranged into HSI datacubes with each λ plane along the z-axis representing the (x,y) intensity values measured for a particular wavelength.

Since the scanning process is integrated within the camera, HSI datacubes of stationary samples can be efficiently created in seconds. Researchers do not have to configure translation stages, conveyor belts or other methods of motion control. On most imaging platforms, one simply replaces a conventional camera with a Snapscan camera to make an imaging system capable of Hyperspectral Imaging. The systems are robust enough for lab operation as well as for use in the field.

Snapshot: introduction

imec Snapshot technology - ideal for high-throughput spectral imaging
If the specific spectral bands of interest are known in advance and higher throughput near-real-time imaging is desirable, there’s no need to capture a broad spectral range. A Snapshot approach permits a tradeoff of spatial resolution for faster acquisition speeds opening new possibilities for near-real-time generating of spectral and spatial information.

Snapshot: Theory of Operation

Snapshot cameras are used by scientists and engineers to spectrally inspect samples by rapidly acquiring selected spectral and spatial data from a field of view at video rates. In Snapshot cameras, the Fabry-Perot band-pass spectral filters are deposited on the image sensors in a 3x3, 4x4 or 5x5 mosaic pattern in a manner that is conceptually similar to that of the familiar RGB Bayer pattern.

The cameras are c-mount compatible, which allows their use in microscopy or with conventional camera lenses. Snapshot cameras generate spectral and spatial data at video rates, capturing the image circle that is projected by the lens or microscope optics onto the focal plane. Pixel values are read out and can now be rearranged with each λ plane along the z-axis representing the spatially sampled (x,y) intensity values measured for a particular wavelength.

Microscopy Example

imec Video

Snapscan Cameras

lmec’s Snapscan [VNIR or SWIR] high resolution hyperspectral camera technology is a major breakthrough for hyperspectral imaging application research. A high quality hypercube data is created within a few seconds with high signal-to-noise ratio and unmatched spatial and spectral resolution. While the Snapscan kit enables application research of the highest quality, it is user-friendly since it does not require an external scanning system. It integrates all key components required: the spectral image sensor, optics, illumination and imec’s hyperspectral imaging software: HSI Snapscan.

RGB rendering and classified image (red blood cells versus white blood cells) from a dataset of 150 spectral images of 7Mpx spatial resolution (snapscan VNIR camera was mounted onto a LEICA microscope for spectral imaging of a blood smear test

Snapscan SWIR camera & lens with a Full Lab setup

Snapscan selection table

Spectral range	Spectral resolution	Spatial resolution	FWHM	Interface	Optical Interface
470 – 900 nm (VNIR)	150 bands	up to 3600 x 2048px (7Mpx per band)	~ 10 – 15 nm (collimated)	USB3.0 + GPIO for triggering	C-Mount
1100 – 1700 nm (SWIR)	100 bands (SWIR)	up to 1200 x 640 px (0.8MP RAW per band)	~ 5 – 10 nm (collimated)	USB3.0 + GPIO for triggering	C-Mount

Click on a green button within the above table to get more details about a particular camera model.

Snapshot Cameras

lmec’s Snapshot [VIS, RedNIR, NIR or SWIR] range spectral imaging cameras offer a simple, fast and easy solution for the video-rate acquisition of spectral images of sample materials and their subsequent analysis.

Snapshot cameras are robust enough for lab operation as well as for use in the field, on fixed or movable platforms including UAVs.

Videos: selected applications of imec's Snapshot cameras

The video above shows the application of Snapshot cameras in generating real-time oxygenation maps during surgery. A Snapshot camera with a 4x4 mosaic was used to reconstruct Hb/HbO/HBT maps from images taken during brain surgery on an epileptic patient. Images are courtesy of Polytechnique Montreal.

The video above shows that Snapshot cameras can be used for spectral imaging at video rates. As the leaf is moved in the video, spectral data is acquired and analyzed in real-time. The technology represented in Snapshot cameras is ideally suited for applications in which spectral discrimination and classification must be performed at video rate. While there is a tradeoff in spatial and spectral resolution as compared with Snapscan technology, Snapshot is the technology of choice for spectral imaging applications requiring high-throughput or for imaging from moving platforms, including UAVs.

Snapshot selection table

Spectral resolution

Spatial resolution

Base imager type

Acquisition speed

Optics

Host Interface

16 bands in 460 – 600 nm range (VIS)

2048x1088 (2MP after reconstruction)

AMS CMV2000 CMOS detector

up to 120 spectral cubes per second

C-mount

Camera models with either USB3.0 or GigE are available

15 bands in 600 – 860 nm range (RedNIR)

24 bands in 665 – 960 nm range (NIR)

9 bands in 1100 – 1650 nm range (SWIR)

640 x 512 (raw sensor resolution)

InGaAs based, Cardinal 640 sensor with TEC cooler

USB3.0