Artificial vision used to detect rotten oranges and pick through mandarins
By Ben Coxworth
October 17, 2011
There's a reason that the oranges you see in the store usually aren't rotten - someone at a sorting facility has already looked over all the oranges coming in from the fields, and taken out the spoiled ones. This is typically done with the help of ultraviolet light, which illuminates the essential oils in the rinds of rotten oranges. Such an approach is subject to human error, however, plus workers can only remain in the vicinity of the harmful UV light for limited periods of time. Now, scientists from Spain's Valencian Institute of Agrarian Research (IVIA) have created a machine that does the same job automatically. While they were at it, they also came up with one that sorts oranges according to aesthetic appeal, and one that sorts mandarin segments.
All of the machines incorporate artificial vision technology, which allows computers to interpret data gathered by linked video cameras.
The rotten-orange-detecting machine uses the traditional UV light to detect the offending fruits, then mechanically removes them from the production line. Another machine, at the rate of 15-20 fruits per second, uses visible light to sort citrus fruits according to the quality of their appearance. Based on factors such as color and skin damage, fruits that are still perfectly edible are routed either to high-end retailers, or to markets where looks aren't so important. It's reminiscent of a machine recently built at Montreal's McGill University, that uses visible and near-infrared light to visually inspect and classify cuts of meat.
The mandarin-sorting machine separates mandarin segments from one another by placing them on a vibrating platform, then carries them on a conveyor belt to an inspection area. There, at the rate of 28 segments per second, they are inspected and sorted according to whether or not they are complete segments, or if they contain pips. Foreign bodies such as skin are also identified and removed.
Yet another machine, which is about the size of a large tractor, can be used to sort oranges by quality while still in the field. Workers place picked oranges on its conveyor belt, which brings the fruits into the machine for inspection. In this way, the fruit will come in from the fields already sorted and ready to go.
The scientists are also looking into the use of MRI, CAT scans and X-rays, along with hyperspectral imaging. This last technology "collects and processes information from a large part of the electromagnetic spectrum and provides individual spectral measurements for each pixel," according to IVIA. It could be used to identify the concentrations of chemical compounds that change as fruit ripens or rots.
A report on the research was recently published in the journal Food and Bioprocess Technology.