Applying technology to the question of grape quality

10 Nov 2017
Previous  | Next News

This project is supported by funding from the Australian Government Department of Agriculture and Water Resources as part of its Rural R&D for Profit program.

Digital technology is rapidly expanding its presence and impact in the wine sector, and the next stop could be the weighbridge.

Researchers at the Australian Wine Research Institute (AWRI) are trialling a new way to objectively measure bunch rot and matter other than grapes (MOG) in fruit as it arrives using near infrared (NIR) hyperspectral imaging.

The technology already exists and is widely used in the food industry to test for disease and other quality parameters, such as moisture, oil or protein levels in wheat. It works because all organic things have a specific spectral signature.

>
Photo: Australian Wine Research Institute
Hyperspectral camera and lighting system

The aim of the current project, supported by Wine Australia and the Australian Government Department of Agriculture and Water Resources as part of its Rural R&D for Profit program, is to validate the technology in a wine setting and develop a cost-effective approach that would meet the specific needs of winemakers and grapegrowers.

‘There’s a need, especially in large wineries with a huge intake of fruit, to accurately assess the fruit as it comes over the weighbridge’, said Senior Research Scientist Dr Bob Dambergs.

‘You might have inspected the vineyard three to four days before, but in conditions like we had last vintage – when botrytis spread rapidly – things could have changed dramatically.’

The first stage of the work was carried out in the lab, using infected grapes that Dr Dambergs and Dr Paul Petrie had created themselves using table grapes and botrytis spores. The aim was not just to test the imaging results against for traditional ‘wet chemistry’, but also how to tailor the equipment’s potential.

‘The advantage of the lab trial was that it let us know the fingerprint we need to look at’, Dr Dambergs said. ‘If each pixel has a spectrum of 300 data points in it, it actually adds up to quite a huge file in the end.

‘The initial work we’ve done is with full spectral scans and we know that we can pick up infected fruit with that wavelength range, but with those scans we can also start narrowing it down to certain wavelengths so we can make a system that’s cheaper and faster and needs less data storage capacity.’

- Dr Bob Dambergs, Senior Research Scientist, AWRI.

‘If you’re measuring it at the weighbridge you can’t take half an hour to do a test.’

The next stage is to take the technology out to a weighbridge – or at least a simulated weighbridge setting – to test the logistics.

‘We need to get a better understanding of the opportunities and the restrictions of sampling on site’, Dr Petrie said. ‘Would it be feasible to have a camera mounted over the bin to take the image or are there going to be issues with how far the camera is from the bin? Is the resolution high enough and can we get the right lighting to get the spectral signature that we need?’

‘Can we manage logistical problems or are we going to have to have a solution where we take a sample out of the bin and put it into a lab environment right next to weighbridge?’

Dr Petrie hopes that by the end of next vintage they will be in a situation where a company could make a decision about using the technology and ‘we could provide them with enough information so that they could implement it’.

‘We’re not here to generate an industrial solution, but we need to go far enough that we could inform a solution.’

The weighbridge trials will also test the potential to streamline the imaging and analysis process by developing software that can both control the camera and analyse the data it is collecting.

>Hyperspectral imaging of botrytis infected grapes
Photo: Australian Wine Research Institute
Hyperspectral imaging of botrytis infected grapes: on the left are RGB images of grapes, on the right are the same grapes with false colour overlays derived from hyperspectral image analysis.