Hyperspectral system trade-offs for illumination, hardware and analysis methods: a case study of seed mix ingredient discrimination

The discrimination power of a hyperspectral imaging system for image segmentation or object detection is determined by the illumination, the camera spatial–spectral resolution, and both the pre-processing and analysis methods used for image processing. In this study, we methodically reviewed the alternatives for each of those factors for a case study from the food industry to provide guidance in the construction and configuration of hyperspectral imaging systems in the visible near infrared range for food quality inspection. We investigated both halogen- and LED-based illuminations and considered cameras with different spatial–spectral resolution trade-offs. At the level of the data analysis, we evaluated the impact of binning, median filtering and bilateral filtering as pre- or post-processing and compared pixel-based classifiers with convolutional neural networks for a challenging application in the food industry, namely ingredient identification in a flour–seed mix. Starting from a basic configuration and by modifying the combination of system aspects we were able to increase the mean accuracy by at least 25 %. In addition, different trade-offs in performance-complexity were identified for different combinations of system parameters, allowing adaptation to diverse application requirements.

Download Full-text

Determination of Basic Density and Moisture Content of Loblolly Pine Wood Disks Using a near Infrared Hyperspectral Imaging System

Journal of Near Infrared Spectroscopy ◽

10.1255/jnirs.948 ◽

2011 ◽

Vol 19 (5) ◽

pp. 401-409 ◽

Cited By ~ 26

Author(s):

Christian R. Mora ◽

Laurence R. Schimleck ◽

Seung-Chul Yoon ◽

Chi N. Thai

Keyword(s):

Moisture Content ◽

Hyperspectral Imaging ◽

Loblolly Pine ◽

Near Infrared ◽

Imaging System ◽

Basic Density ◽

Pine Wood

Download Full-text

Rapid ripening stage classification and dry matter prediction of durian pulp using a pushbroom near infrared hyperspectral imaging system

Measurement ◽

10.1016/j.measurement.2021.110464 ◽

2021 ◽

pp. 110464

Author(s):

Sneha Sharma ◽

K.C. Sumesh ◽

Panmanas Sirisomboon

Keyword(s):

Hyperspectral Imaging ◽

Dry Matter ◽

Near Infrared ◽

Imaging System ◽

Ripening Stage ◽

Stage Classification

Download Full-text

Prediction of Douglas-Fir Lumber Properties: Comparison between a Benchtop Near-Infrared Spectrometer and Hyperspectral Imaging System

Applied Sciences ◽

10.3390/app8122602 ◽

2018 ◽

Vol 8 (12) ◽

pp. 2602 ◽

Cited By ~ 4

Author(s):

Laurence Schimleck ◽

Joseph Dahlen ◽

Seung-Chul Yoon ◽

Kurt Lawrence ◽

Paul Jones

Keyword(s):

Hyperspectral Imaging ◽

Near Infrared ◽

Imaging System ◽

Nir Spectroscopy ◽

Physical And Mechanical Properties ◽

Douglas Fir ◽

Modulus Of Rupture ◽

Visual Grade ◽

Fit Statistics ◽

Infrared Spectrometer

Near-infrared (NIR) spectroscopy and NIR hyperspectral imaging (NIR-HSI) were compared for the rapid estimation of physical and mechanical properties of No. 2 visual grade 2 × 4 (38.1 mm by 88.9 mm) Douglas-fir structural lumber. In total, 390 lumber samples were acquired from four mills in North America and destructively tested through bending. From each piece of lumber, a 25-mm length block was cut to collect diffuse reflectance NIR spectra and hyperspectral images. Calibrations for the specific gravity (SG) of both the lumber (SGlumber) and 25-mm block (SGblock) and the lumber modulus of elasticity (MOE) and modulus of rupture (MOR) were created using partial least squares (PLS) regression and their performance checked with a prediction set. The strongest calibrations were based on NIR spectra; however, the NIR-HSI data provided stronger predictions for all properties. In terms of fit statistics, SGblock gave the best results, followed by SGlumber, MOE, and MOR. The NIR-HSI SGlumber, MOE, and MOR calibrations were used to predict these properties for each pixel across the transverse surface of the scanned samples, allowing SG, MOE, and MOR variation within and among rings to be observed.

Download Full-text

Chemometric Tools for NIRS and NIR Hyperspectral Imaging

Bulletin of University of Agricultural Sciences and Veterinary Medicine Cluj-Napoca Agriculture ◽

10.15835/buasvmcn-agr:8658 ◽

2012 ◽

Vol 69 (1) ◽

Author(s):

Laura M. DALE ◽

André THEWIS ◽

Ioan ROTAR ◽

Juan A. FERNANDEZ PIERNA ◽

Christelle BOUDRY ◽

...

Keyword(s):

Hyperspectral Imaging ◽

Near Infrared ◽

Agricultural Sector ◽

Spatial Information ◽

Imaging System ◽

Agricultural Products ◽

Chemical Information ◽

Special Focus ◽

Large Area ◽

Analytical Tools

Nowadays in agriculture, new analytical tools based on spectroscopic technologies are developed. Near Infrared Spectroscopy (NIRS) is a well known technology in the agricultural sector allowing the acquisition of chemical information from the samples with a large number of advantages, such as: easy to use tool, fast and simultaneous analysis of several components, non-polluting, noninvasive and non destructive technology, and possibility of online or field implementation. Recently, NIRS system was combined with imaging technologies creating the Near Infrared Hyperspectral Imaging system (NIR-HSI). This technology provides simultaneously spectral and spatial information from an object. The main differences between NIR-HSI and NIRS is that many spectra can be recorded simultaneously from a large area of an object with the former while with NIRS only one spectrum was recorded for analysis on a small area. In this work, both technologies are presented with special focus on the main spectrum and images analysis methods. Several qualitative and quantitative applications of NIRS and NIR-HSI in agricultural products are listed. Developments of NIRS and NIR-HSI will enhance progress in the field of agriculture by providing high quality and safe agricultural products, better plant and grain selection techniques orÂ compound feed industryâ€™s productivity among others.

Download Full-text

3D Plant Modelling Using Spectral Data from Visible to Near Infrared Range

Advances in Environmental Engineering and Green Technologies - Computer Vision and Pattern Recognition in Environmental Informatics ◽

10.4018/978-1-4666-9435-4.ch013 ◽

2015 ◽

pp. 273-294

Author(s):

Ali Zia ◽

Jie Liang

Keyword(s):

Hyperspectral Imaging ◽

Material Properties ◽

Near Infrared ◽

Infrared Range ◽

Plant Surface ◽

Imaging Device ◽

Plant Phenomics ◽

Physical Traits ◽

Plant Modelling ◽

Near Infrared Range

Plant phenomics research requires different types of sensors employed to measure the physical traits of plant surface and to estimate the biomass. Of particular interests is the hyperspectral imaging device which captures wavelength indexed band images that characterize material properties of objects under study. This chapter introduces a proof of concept research that builds 3D plant model directly from hyperspectral images captured in a controlled lab environment. The method presented in this chapter allows fine structural-spectral information of an object be captured and integrated into the 3D model, which can be used to support further research and applications. The hyperspectral imaging has shown clear advantages in segmenting plant from its background and is very promising in generating comprehensive 3D plant models.

Download Full-text

A Near-Infrared Hyperspectral Imaging System for Monitoring Wound Healing Post Amputation

2007 IEEE Dallas Engineering in Medicine and Biology Workshop ◽

10.1109/embsw.2007.4454160 ◽

2007 ◽

Cited By ~ 1

Author(s):

Karel J. Zuzak ◽

Santosh Hariharan ◽

Sabira C. Naik ◽

Manasi Mandhale ◽

Edward H. Livingston

Keyword(s):

Wound Healing ◽

Hyperspectral Imaging ◽

Near Infrared ◽

Imaging System

Download Full-text

Preliminary Design and Performance of Counter Rotating Turbines for Open Rotors: Part II — 0-D Methodology and Case Study for a 160 PAX Aircraft

Volume 1: Aircraft Engine; Fans and Blowers; Marine ◽

10.1115/gt2016-57921 ◽

2016 ◽

Author(s):

Pablo Bellocq ◽

Inaki Garmendia ◽

Jordane Legrand ◽

Vishal Sethi

Keyword(s):

Design Methodology ◽

Design Space Exploration ◽

Design Space ◽

Space Exploration ◽

Preliminary Design ◽

Regulatory Constraints ◽

Trade Offs ◽

And Performance ◽

The Impact

Direct Drive Open Rotors (DDORs) have the potential to significantly reduce fuel consumption and emissions relative to conventional turbofans. However, this engine architecture presents many design and operational challenges both at engine and aircraft level. At preliminary design stages, a broad design space exploration is required to identify potential optimum design regions and to understand the main trade offs of this novel engine architecture. These assessments may also aid the development process when compromises need to be performed as a consequence of design, operational or regulatory constraints. Design space exploration assessments are done with 0-D or 1-D models for computational purposes. These simplified 0-D and 1-D models have to capture the impact of the independent variation of the main design and control variables of the engine. Historically, it appears that for preliminary design studies of DDORs, Counter Rotating Turbines (CRTs) have been modelled as conventional turbines and therefore it was not possible to assess the impact of the variation of the number of stages (Nb) of the CRT and rotational speed of the propellers. Additionally, no preliminary design methodology for CRTs was found in the public domain. Part I of this two-part publication proposes a 1-D preliminary design methodology for DDOR CRTs which allows an independent definition of both parts of the CRT. A method for calculating the off-design performance of a known CRT design is also described. In Part II, a 0-D design point efficiency calculation for CRTs is proposed and verified with the 1-D methods. The 1-D and 0-D CRT models were used in an engine control and design space exploration case study of a DDOR with a 4.26m diameter an 10% clipped propeller for a 160 PAX aircraft. For this application: • the design and performance of a 20 stage CRT rotating at 860 rpm (both drums) obtained with the 1-D methods is presented. • differently from geared open rotors, negligible cruise fuel savings can be achieved by an advanced propeller control. • for rotational speeds between 750 and 880 rpm (relatively low speeds for reduced noise), 22 and 20 stages CRTs are required. • engine weight can be kept constant for different design rotational speeds by using the minimum required Nb. • for any target engine weight, TOC and cruise SFC are reduced by reducing the rotational speeds and increasing Nb (also favourable for reducing CRP noise). However additional CRT stages increase engine drag, mechanical complexity and cost.

Download Full-text

Hyperspectral Imaging for Predicting Soluble Solid Content of Starfruit

Jurnal Teknologi ◽

10.11113/jt.v73.3480 ◽

2015 ◽

Vol 73 (1) ◽

Cited By ~ 1

Author(s):

Feri Candra ◽

Syed Abd. Rahman Abu Bakar

Keyword(s):

Hyperspectral Imaging ◽

Near Infrared ◽

Imaging System ◽

Solid Content ◽

Partial Least Square ◽

Least Square ◽

Coefficient Of Determination ◽

Soluble Solid Content ◽

Calibration Model ◽

Soluble Solid

Hyperspectral imaging technology is a powerful tool for non-destructive quality assessment of fruits. The objective of this research was to develop novel calibration model based on hyperspectral imaging to estimate soluble solid content (SSC) of starfruits. A hyperspectral imaging system, which consists of a near infrared camera, a spectrograph V10, a halogen lighting and a conveyor belt system, was used in this study to acquire hyperspectral images of the samples in visible and near infrared (500-1000 nm) regions. Partial least square (PLS) was used to build the model and to find the optimal wavelength. Two different masks were applied for obtaining the spectral data. The optimal wavelengths were evaluated using multi linear regression (MLR). The coefficient of determination (R2) for validation using the model with first mask (M1) and second mask (M2) were 0.82 and 0.80, respectively.

Download Full-text