.: ImSpector standard :.
.: ImSpector enhanced :.
.: flyer(pdf) :.
ImSpector principle
imaging spectroscopy
principleIn every spectrometer the dispersive element is the most important component for the determination of it optical properties because its determines the spectral range, the spectral resolution and the transmission.
The dispersive element separates the light depending on its wavelengths and projects these fractions on different spatial positions. We will denote the combination of a dispersive element an imaging spectrograph if it is together with some optical components (entrance slit, lens system, housing,....) suitable for spectral imaging. One special realization of such and imaging spectrograph which is regarded to be currently the most interesting spectrograph for industrial applications is the imaging spectrograph developed at the VTT in Oulu which is described in the comprehensive work of Mauri Aikio who also gives a short summary over the history of spectral imaging.
Conventional commercial spectrometers or spectrophotometers are usually able to measure only the optical spectrum from a specified surface area as one point. This is done either with one detector scanning the spectrum in narrow wavelength bands or with a diode array detector, in which case all the spectral components are electronically acquired at once. If one desires to measure the spectrum at several spatial locations of the specified surface -- like it is often required in industrial applications -- the target under examination or the measuring instrument has to be mechanically moved.
In Fig.1 a schematic of the imaging spectrograph ImSpector -- developed by Specim Ltd. based on the results of Aikio -- is shown. The dispersive, stationary spectrograph module can be coupled with a lens system or fiber optic system on the target side and should be connected to a matrix detector for a simultaneous mapping of the spectral image data.
Schematic of the imaging spectrograph embedded into typical components
on both the target and the detector side
on both the target and the detector side
The spectrograph can be coupled with a lens system or fiber optic system on the target side and should be connected to a matrix detector for a simultaneous mapping of the spectral data.
One dimension of the detector corresponds to a spatial line image and the other dimension corresponds to the spectrum. The usage of a holographic transmission grating as a realisation of an optical filter has the advantage of a high transmission rate and low aberrations. The combination of this grating with two prism to a so-called prism-grating-prism element allows an optical design with one axis. Especially for non-laboratory applications this one axis design makes the experimental set-up much more easier.
Emerging from the imaging spectrograph is a two dimensional image of the 'objects slice' where one axis represents the spatial width of the image in the direction of the slit and the other axis represents the wavelengths of the spectrum. When this image is typically collected by a matrix detector -- a monochrome CCD or CMOS camera in the visible or near infrared spectral range -- the first axis represents the average of the light coming from a small part of the 'object slice'. The second axis represents the light intensity at a particular wavelength. This data matrix consists of the spatial information in each row and the spectral information in each column. The imaging spectrograph is designed for 2/3 inch matrix detectors with the long axis parallel to the spatial direction and the short axis parallel to the spectral directions. In other words, the spectrograph converts an area (FPA) monochrome detector (camera) to a spectral line imaging system.
With one image capture of the matrix detector in combination with the imaging spectrograph the recorded data corresponds to a single line of the target surface. There maybe still remain the task of scanning the surface in the other dimension perpendicular to this line as a function of time. By using an imaging spectrograph this is much easier to accomplish compared to methods using conventional spectrometers. For scanning the surface only a linear translation of either the object or the detector system is required. In some applications the movement of the object (process stream, web,...) automatically forms the other spatial dimension.
Concerning a spectral imaging system the spectrograph together with the lens system is one of the components which can be essentially denoted as ready for industrial applications. Because of the absence of any moving mechanical or optical components and its rigid metal housing the spectrograph is quite robust. The spatial resolution is in the order of a high resolution camera and the spectral resolution can be compared with the resolution of standard diode array spectrometers. The spectrographs are available in specific versions with several different spectral intervals beginning at a wavelength of 200 nm and currently ending at 2400 nm -- e.g 400-800 nm or 1100-1700 nm. We assume that most of the work has succesfully be done concerning the development of the spectrograph to make this component fit for industrial application.
Color determination using spectral images
The imaging spectrograph offers a wide range of new applications. Despite color measurement with the ImSpector is only a small fragment of its possible application range it is one of the most interesting and difficult applications.
The image shows an example of a target for color measurement is shown. To give a graphical description of the principle of spectral imaging the target does only vary in one spatial direction. The measurement area detected by the imaging spectrograph is located between the two white lines. The geometric length and width of this area is determined by the distance of the entrance slit to the target (here 100 mm). Increasing the distance between the target and the imaging spectrograph the surface of this area increases proportionally. Corresponding to the properties of the optical lens used, this area can vary from microscopic scale (below one mm) to macroscopic scale (more than several 10 m).
The following image shows a falsified color image of the spectral image data recorded with the imaging spectrograph corresponding to the displayed measurement area. It is very important to get used to this special arrangement of spectral data that can be explained as: The spatial direction (horizontal) corresponding to the horizontal spatial direction in. In the spectral direction the wavelengths vary from 320 nm to 800 nm. A color of blue means low intensity and a color of red means high intensity in that spectral range.
The matrix detector used for this measurement has a resolution of 640x512 pixels converted with a 12 bit analog digital converter. This means the spectral image shown before consist of 640 spectra each with 512 spectral bands. In practical applications the high dynamic range and resolution of the analog digital conversion is more important than the number of pixels of the CCD matrix detector. Because of the enormous flexibility of the analysis of the digital spectral imaging data for every application, the focus on special wavelength bands of the spectra related to the region on the matrix detector can be individually adapted.
In Fig.4 the spectral imaging data of Fig.2 is shown as a three dimensional plot which reproduces an illustrated view of the dimensions of spectral imaging. Scanning over the target or recording a moving target leads even to a four dimensional space consisting of time (or one spatial direction), space, wavelength and intensity. Using a sequence of spectral images the image of the original target can be reconstructed to RGB and displayed.
In Fig.5 the spectra corresponding to the color strips of the target are shown in a conventional wavelength versus intensity plot. Each spectrum was calculated by averaging and smoothing all spectra of one color strip.
Of course every standard RGB detector system could distinguish between the different colored strips. But imagine a more complex arrangement of spectral and spatial differences of the samples: then spectroscopic analysis is necessary.