Search

Supplemental Material for Development of a polarized hyperspectral imaging system for investigation of absorption and scattering properties by Jun-Li Xu, Alexia Gobrecht, Nathalie Gorretta, Daphné Héran, Aoife A Gowen and Ryad Bendoula in Journal of Near Infrared Spectroscopy

Hyperspectral (HS) images have the characteristic of containing both spectral and spatial information of a sample. Typically, spectral information can be related with chemical and physical properties through multivariate regression models. The application of these models onto HS images results in prediction maps, which provide an estimation of the modelled chemical information for each pixel of the image. This approach has wide applications in food processing industries for online monitoring of product quality and process control. One of the main difficulties derived from an imaging set up, is that the size of the pixels is usually much smaller than the area required to obtain a wet chemical reference. This means that, as opposed to point spectroscopy, the performance of the estimations cannot be evaluated by directly comparing observed and estimated values for each pixel. Moreover, the selection of regression model parameters, such as the number of latent variables (LV) in a partial least squares (PLS) model, cannot be assessed on a pixel basis either. Nonetheless, compared to point spectroscopy, HS imaging does provide information on the spatial distribution of the predicted values. The objective of this work is to propose a quantitative approach to use spatial information of prediction maps for supporting the evaluation of regression models applied to HS images. This approach is based on the use of geostatistical indexes, which allow decomposing the total variance of the prediction maps into two components: non spatially structured and spatially structured variance, represented respectively by the nugget effect (C0) and the partial sill (C1). This strategy was tested in a simulated dataset and two real case studies. Geostatistical indices of the prediction maps were compared to model performance metrics for PLS models with increasing number of LV. As a result, this work stablishes a connection between linear regression model performance estimates and the spatial decomposition of variance in prediction maps, when the ground truth to estimate is spatially structured. The presented study [1], allowed to evaluate HS imaging models, not only from average estimations which can be compared to reference values, but also from the spatial structure of prediction maps. This approach does not require ground truth values and could be used as a source of information for supporting the choice of optimum calibration options, such as the number of LV, or the pre-treatments, complementing the traditional visual inspection of prediction maps with quantitative and objective metrics. Further works could explore the effect of spatial structure at different scales in the data. This approach could also be applied to evaluate variations of other parameters in the model, such as different methods for spectral pretreatments.

Conferntia Chemetrica 2017, Gyöngyös-Farkasmály, Hungary, 3-6 September 2017 Array detectors improve data collection speed in hyperspectral chemical imaging, yet are prone to striping noise patterns in the image plane, which is difficult to remove. This type of noise affects spectral features and disturbs visual impression of the scene. We found that this type of noise depends on the material composition and setting parameters, ie, pixel size, and it also varies in accordance with the signal intensity of the observed wavelength. To address this, we proposed a new correction method on the basis of the application of chemometric techniques in the image plane of each wavelength. To verify the effectiveness of this method, infrared transmission images of the 2″ × 2″ positive, 1951 USAF Hi‐Resolution Target and biomaterial samples were obtained with a 16‐element (8 × 2) pixel array detector. Point detector images of some samples were also acquired and used as reference images. The proposed correction method produced substantial improvements in the visual impression of intensity images. Principal components analysis was performed to inspect spectral changes after preprocessing, and the results suggested that the major spectral features were not altered while the stripes on intensity images were removed. Spectral profiles and principal components analysis loadings inspection confirmed the smoothing ability of this correction method. As traditional preprocessing techniques, standard normal variate and derivative transformation were not able to remove line artifacts, especially on the biomaterial images. Overall, the proposed method was effective for removing striping noise patterns from infrared images with a minimal alteration of the valuable hyperspectral image information. European Commission - Seventh Framework Programme (FP7) University College Dublin China Scholarship Council

Mealiness is known as an important internal quality attribute of fruits/vegetables, which has significant influence on consumer purchasing decisions. Mealiness has been a topic of research interest over the past several decades. A number of destructive and nondestructive techniques are introduced for mealiness detection. Nondestructive methods are more interesting because they are rapid, noninvasive, and suitable for real-time purposes. In this review, the concept of mealiness is presented for potato, apple, and peach, followed by an in-depth discussion about applications of destructive and nondestructive techniques developed for mealiness detection. The results suggest the potential of electromagnetic-based techniques for nondestructive mealiness evaluation. Further investigations are in progress to find more appropriate nondestructive techniques as well as cost and performance.