Close this search box.
Close this search box.

Cayenne and Oregano

Quality Inspection of Herbs and Spices

George Killian, Headwall Applications Scientist, adjusts the lens on an MV.C NIR mounted on the perClass Mira Stage while it scans herbs and spices
Key: blue = oregano, magenta = adulterant

Herbs and spices are one of the rockstars of food fraud. Their complex cross-border supply chains, high price per kilogram, and the fact that they are often sold in powder or particulate forms make them prime targets for adulteration, dilution, and substitution with cheaper materials.1

Oregano is often mixed with other, similar-looking but cheaper ingredients like olive leaf and myrtle. Studies conducted in recent years on ground oregano sold in stores showed that
nearly a quarter of the samples contained non-oregano ingredients. The scale of adulteration was found to range from
30% to 70% by weight.2

Adulteration of oregano can be detected using liquid chromatography-mass spectrometry (LC/MS) analysis, and purity of spices (such as red pepper) can be determined by gas chromatography, but these methods require significant time and effort in sample preparation. Optical microscopy is also used, but identification and counting of individual small particles is tedious.

A hyperspectral imaging (HSI) system paired with a machine learning algorithm can quickly show the user if non-oregano ingredients are present in the sample and help quantify them in real-time.



In a test recently conducted at Headwall, samples of oregano, French thyme, and myrtle were obtained and scanned with the MV.C NIR (range 900 to 1700 nm) sensor on the perClass Mira® Stage and captured in the perClass Mira Software. The scans were used as a reference for spectral classification method development, with perClass Mira’s machine learning software.

An oregano class was created by selecting known oregano leaf pixels to the classification model. Upon investigation of the oregano image, there were regions of leaves that had significantly different spectra than the oregano leaves, as shown below.

A small sampling of these pixels were added to an adulterant class, and a new classification model was employed. The resulting false-colored RGB classification image showed multiple regions of non-oregano adulterant leaves in the sample. These adulterants are impossible to determine by the naked eye, or even with high resolution RGB. But with HSI, the adulterants are detected.

This model was further improved by adding spectral indexes for French thyme and common myrtle leaves, two common adulterants. A mixture of oregano and these common additives were classified with the perClass Mira software, and these adulterants were shown in the software.

The perClass Mira software allows for further processing of the image, and provides the user with a workflow to get a percentage of adulterant pixels in a sample of oregano to produce a quantified result and make quality decisions in seconds.




When it comes to ground spices, there are many more types of additives that can be added to mask the true grade of the product. Dyes, chalk powders, clays, and seed powders are often added to red pepper powder as adulterants.3

In an internal study conducted by Headwall Photonics, pure red pepper and red clay were gathered, and combined in different mass percent concentrations. These powders were then scanned with a MV.C NIR sensor on a perClass Scanning Stage, and the data gathered and analyzed in perClass Mira Software. The machine learning software was trained on the known concentrations, and able to estimate the concentrations of red pepper test samples to a high degree, as shown in the regression model below.

With an inline or quality test stand setup, a perClass Mira model could be trained and deployed for real-time estimation of contaminant concentration in different spice powders.

A side-by-side comparison of an RGB image of oregano, next to a hyperspectral false-color image of the same oregano.
Left: Simulated RGB color rendition of a hyperspectral image of oregano leaves. Right: False-color classified hyperspectral image of the same oregano leaves, showing regions with different spectral characteristics.
Regression model of two different samples of cayenne pepper, Sample A is 96% cayenne and Sample B is 98%
The regression model results from a hyperspectral scan of cayenne pepper, showing the estimated mass percent of cayenne compared to the measured mass percent of cayenne
Graph showing the accuracy of a regression model created using perClass Mira Software to estimate the concentrations of red pepper when combined with various adulterants.


Unlike LC/MS, FTIR, and other laboratory techniques, operating Headwall’s hyperspectral imaging systems require no specialist knowledge. A simplified user interface can be configured in perClass Mira base on individual requirements, making the HSI system suitable for operation in a QC environment.

Spectral classification via the perClass Mira Software makes the task of assessing herb and spice quality less challenging by providing a clear indication of the presence of adulterants, and also provides a workflow of detecting unknown adulterants and additives.

In the past, significant spectroscopic expertise was needed to determine the quality of herbs and spices, via either destructive or spectral classification methods. This is no longer the case with advancements in HSI and perClass Mira’s machine learning software. The intuitive software of perClass Mira paired with the innovative MV.C NIR has simplified this process, and makes this exciting technology widely accessible.



More Application Notes

An RGB image of a mixture of herbs and spices alongside a hyperspectral false-color scan of the same pile.
Cayenne and Oregano

Spectral classification via the perClass Mira Software makes the task of assessing herb and spice quality less challenging by providing a clear indication of the

Read More »

Explore more content

Hyperspectral 101

Webinar / Videos

Published Research