ImSpectorThe series imaging spectrometer is a high-performance spectrometer launched by Specim, a global leader in hyperspectral imaging technology. It is designed specifically for VIS (380-800nm), VNIR (400-1000nm), and NIR (900-1700nm) bands. The ImSpector imaging spectrometer provides a simple, high-performance, and cost-effective integration method for integrators and machine manufacturers around the world. When combined with scientific grayscale CCD/CMOS cameras or InGaAs sensors, it forms a line scan spectroscopic imaging device that is applied to various inspection, classification, and other machine vision solutions for daily use.

ImSpectorThe imaging spectrometer optimizes the spectral resolution, detector size, spatial resolution, and imaging speed of each module, providing the highest optical performance distortion free images on the market to meet the most demanding application requirements.

Optional front optical lens:

ü Standard series: OL8, OL12, OL17, OL23, OL35 for 2/3 inch or smaller detectors

ü Enhanced series: OLE9, OLE18.5, OLE23, OLE140 for 2/3 inch or larger detectors

ü Other series: OLES15, OLES22.5, OLES30, OLES56 for N17E

Optional accessories:

ü Mechanical Shutter (Enhanced Series)

ü Collecting optical fibers

ü Bandstop filter, OBF 570 (rectangular 14 × 12mm or circular 20mm Ø and 17mm Ø), used for V10 and V10E

ü Fiber optic diffuse irradiance sensor FODIS (Enhanced Series) for light source monitoring

Technical parameters:

Note:

*1 Can install a band stop filter in front of the detector window

*2 The spectral and spatial resolution of the system also depend on the discrete imaging characteristics of the detector and the quality of the lens

Application Case 1: Classification of Peanuts Naturally Polluted by Aflatoxin B1

Xueming He and other researchers from the School of Food Science and Engineering at Nanjing University of Finance and Economics used an ImSpector V10e spectrometer and EMCCD camera to form a 400-1000nm hyperspectral imaging system. They extracted and integrated spectral, color, and texture features, and measured the reference AFB1 level using enzyme-linked immunosorbent assay (ELISA) to achieve a non-destructive hyperspectral imaging method for distinguishing between normal and natural aflatoxin B1 (AFB1) contaminated peanuts.

Figure 1-1: Schematic diagram of hyperspectral imaging system (left); Peanut sample RGB and segmentation processing images (right): (a1) - (a4) are the pre segmentation RGB image, ROI binary image, segmented RGB image, and segmented grayscale image of the peanut with the lowest AFB1 content (0.1 ppb) in sequence; (b1) - (b4) are the corresponding images of peanuts with the highest AFB1 content (599.21 ppb)

Different preprocessing methods were applied to the entire spectrum, and linear discriminant analysis (LDA) results showed that Savitzky Golay smoothing (SGS) followed by standard normal transformation (SNV) can achieve optimal discrimination, with accuracies of 90% and 92% for the calibration and validation sets, respectively. Finally, the performance of Partial Least Squares Discriminant Analysis (PLS-DA) and Support Vector Machine (SVM) was compared with LDA, and the SVM with RBF kernel showed the best accuracy of 93% and 94% for the calibration and validation sets, respectively.

Figure 1-2: (a) Original spectra of all 150 peanut samples and (b) SGS+SNV spectra

This study demonstrates the potential application of hyperspectral imaging in direct classification of peanut AFB1 pollution and demonstrates that the combination of texture and spectral features can improve modeling results.

Application Case 2: Non destructive Rapid Variety Identification and Visual Expression of Grape Seeds

Researchers such as Yong He from the School of Biosystems Engineering and Food Science at Zhejiang University used the ImSpector N17E spectrometer+ Xeva 992The camera formed the HSI system, which collected hyperspectral images of 14015, 14300, and 15042 grape seeds from three grape varieties in the 874-1734nm spectral range. Preprocess pixel level spectra through wavelet transform, and then extract the spectra of each grape seed. Perform principal component analysis (PCA) on hyperspectral images, using the scores of the first six PCs to qualitatively identify patterns between different varieties, and the loadings of the first six PCs to identify effective wavelengths (EWs).

Figure 2-1: Left: Scoring the images of the first six principal components (PCs): （a）PC1； （b） PC2； （c） PC3； （d） PC4； （e） PC5； And (f) PC6. Right: Load of the first six principal components: （a）PC1； （b） PC2； （c） PC3； （d） PC4； （e） PC5； And (f) PC6

Establish a spectral discrimination model based on EWs using support vector machine (SVM). The results showed that the method can accurately identify the variety of each grape seed, with a validation accuracy of 94.3% and a prediction accuracy of 88.7%. Use external validation images of each variety to evaluate the proposed model and form a classification map, where each individual grape seed is correctly identified as belonging to a different variety.

Figure 2-2: (a) - (f) Original grayscale images and corresponding classification maps of varieties I-III based on this

The overall results indicate that hyperspectral imaging (HSI) technology combined with multivariate analysis can be an effective tool for non-destructive and rapid variety identification and visualization expression of grape seeds. This method has great potential in developing multispectral imaging systems for practical applications.

reference:

[1] He X , Yan C , Jiang X , et al. Classification of aflatoxin B1 naturally contaminated peanut using visible and near-infrared hyperspectral imaging by integrating spectral and texture features[J]. Infrared Physics & Technology, 2021:103652.

[2] Yiying Z , Chu Z , Susu Z , et al. Non-Destructive and Rapid Variety Discrimination and Visualization of Single Grape Seed Using Near-Infrared Hyperspectral Imaging Technique and Multivariate Analysis[J]. Molecules, 2018, 23(6):1352-.