03 November 2014

Matt Philp

Engineering Insight: Pick of the Bunch

  • A dissected onion is illuminated with a white light; the dark regions on the left indicate rot. The project challenge is to see such rot without cutting the onion in half. Photo: Richard Oliver/Plant & Food Research.
  • A whole onion illuminated in the same way shows skin patterns clearly, but no other internal information is obvious. Photo: Richard Oliver/Plant & Food Research.
  • Laboratory set up of a multispectral imaging system to determine fruit quality parameters – camera, lasers, laser optics, and optical shutters. Photo: Philip Rowe.

How can engineers help exporters be certain the fruit they’re exporting is of top export quality? Thanks to the introduction of diagnostic technologies, it has become possible to “see” inside a piece of fruit.

It’s the scenario any New Zealand fruit exporter dreads. A consignment of apples arrives at its overseas destination, where a small sample is cut open for inspection, revealing a rotten core. What left these shores looking like a million dollars turns out to be not-t-for-sale and has to be dumped.

That old exporting horror story is less common today thanks to the introduction of diagnostic technologies to the process of sorting fruit. It has become possible to “see” inside a piece of fruit for defects, internal flesh colour and sugar content without first having to cut the thing open. A joint research exercise involving Plant & Food Research and the University of Waikato’s School of Engineering is seeking to improve that ability by refining diagnostic techniques borrowed directly from the medical imaging field. Instead of revealing brain function anomalies or tumours, these diagnostic tools might be used to reveal fruit firmness, say, or the presence of internal rots.

Plant & Food’s Andrew McGlone is leading the Ruakura-based project. “We fire a beam of light and it interacts with the fruit. Principally it’s a process of light scattering where it comes across a cell, or cellular component and is scattered by the boundary – but there is also absorption and that’s critical, because that is what carries your main composition information. We use a sensor to pick up the light when it exits the fruit, and that sensor tells us something about where the light has been and what it’s encountered on the way through.

“The medical physicists are developing these techniques for non-invasive scanning purposes, such as identifying tumours. We’re simply applying it to fruit and vegetables, the key difference being that we have to scan at a rate of 10 fruit per second, which is the speed of a typical grading machine.”

It’s a big challenge, but the project team is not reinventing the wheel here. One of the industry partners involved is Auckland-based Taste Technologies, which 15 years ago became the first company outside of Japan to commercialise existing near-infrared (NIR) technology for the agriculture industry.

In the medical world, NIR spectroscopy relies on the fact that transmitting and absorbing near infrared light in the human body provides information about changes in blood haemoglobin concentrations. In relation to the brain, for example, where changes in concentrations are associated with neural activity, the technology can provide an alternative to magnetic resonance imaging for non-invasive assessments of brain function. Based on the same concept, Taste Technologies has developed NIR spectroscopy capable of scanning a dozen different types of fruit and vegetable for a raft of flaws.

“We’re able to take bins, clean them up and get a saleable product out the other end,” Business Manager Richard Kestle says.

“We know what a good fruit looks like, what a bad one looks like, because as the light passes through it affects the electromagnetic spectrum differently,” he explains. “We pass light through the fruit and every chemical compound affects the light differently. We’re not looking at whether it’s brown inside or ugly, we’re looking for a physiological change. We can look at an apple and tell if it’s soft, because that softness is the result of a natural chemical breakdown. For the kiwifruit industry, we can look at the gold fruit and can tell if it’s reached the gold stage yet or is still immature, as well as its sugar content. With avocados we look for dry matter, which is an indicator of maturity, and say whether the fruit is good to sell now or not. There’s a range of applications.”

There are shortcomings to the technology, however, which is where Plant & Food and the University of Waikato come into the picture. “At the moment the sensor system largely looks at a whole piece of fruit at once to see if there are any defects,” Dr McGlone comments. “We’re looking at a new technique that is more sensitive to picking up localised information, such as defects hiding in only certain parts of the fruit.”

Onion rot is a particular focus. “It tends to start at the stem end of an onion and can be difficult to find when you’re trying to light up the whole onion. The signal from that particular part is very much diluted – so we’re looking at a technique to harvest the signal by localising the light and the sensor at those regions, while also measuring the other regions, then using the comparison to determine if there is any rot.”

It’s no easy task. The University’s School of Science and Engineering’s Associate Professor, Rainer Kunnemeyer, who is chief advisor to the project, says one of the most vexing issues is how fruit can be scanned in this more focused way while also working at the speed of a typical packhouse grading machine.

“For some applications, such as measuring sugar content, that has already been solved,” he says. “For other properties such as firmness, however, we are still struggling. When the fruit is whizzing past your measurement system at the rate of 10 pieces of fruit per second, you just don’t have time to accumulate enough light to make a proper measurement. One possibility is to use more intense light sources and increase the sensitivity of your detectors.”

Dr Kunnemeyer explains the light in question is non-ionising and entirely benign. “We use light or radiation that is closely related to visible light and that is not harmful in the sense of X-rays or gamma rays. The only issue might be that if you use high intensity radiation and the fruit is exposed for too long it could burn, but the system will ensure against that.”

Besides the speed question, there is a challenge posed by inevitable variation within and between fruit. An avocado, for example, has a stone at its heart and a thick black skin, which makes scanning it an entirely different proposition from scanning a grape.

“Even within individual types of fruit – apples, say – the variation is remarkable,” Dr McGlone says. “There are different sizes and shapes, things such as blemishes on the surfaces – all of which can annoy our readings. We have to devise a system that is robust enough to handle these problems as they present themselves.”

What’s the answer? “There are various tricks available in the toolkit. One of the most obvious solutions is to have two different sources with different characteristics shining at the same spot. One might be more sensitive to the information that you’re looking for, but is also very susceptible to, for example, shape and size effects, whereas the other source might not be so sensitive to some of the complicating factors and can be used to ‘normalise’ the effect.”

Meantime, the researchers have been able to apply the same non-invasive technology used for fruit to test the quality of eggs, meat and milk.

“With eggs we can identify the size of the air bubble in an egg; with milk it’s about discerning the amount of fat and protein,” Dr Kunnemeyer says. “We’re also looking at handheld instruments that could be used by the grower as a tool for assessing fruit on the vine. Is this ready for picking? For breeding purposes, are these grapes better than those others? The challenge with all of these things is to make the solution reliable, light and affordable.

“Of course, the medical research people can throw much more money at developing instruments. My philosophy has always been to have a look at what is viable in the medical field and ask, ‘How do we make this simpler and cheaper to apply to the agricultural field?’.”

Currently, the Ruakura researchers are doing measurements in the laboratory setting with single fruit – statically, in other words. Assuming their line of enquiry “bears fruit” the next step will be to build a prototype and mock up a system to operate at high speed.

The immediate beneficiaries of any development of scanning technology will be Taste Technologies and fellow industry partner Compac, who manufacture fruit grading equipment for the industry. “But it also has potential to help further down the value chain,” Dr Kunnemeyer says. “If you look at ripeness or firmness, this information lets you know more exactly when to put produce onto the market. In theory, the technology could even be used in supermarkets.”

Now there’s a scenario to savour.