Machine vision tests ripeness of fruit
Azor (Israel) – Dates are an important part of the Mediterranean diet. Available in many varieties, one particular favorite is the especially large, meaty Madjoul variety. Dates are grown predominantly in the Middle East, where they are also processed and consumed.
As with all fruit, certain characteristics, such as size, weight, color, and damage to the outside or inside of the fruit determine the product quality and price. Alongside the size and weight, color and surface characteristics play an important role in marketing dates. It is not unusual for the skin to separate from the meat of the fruit and, contingent upon humidity combined with the sugar content, for blisters to form on the date’s surface. These blisters have an effect not only on the taste; for aesethetic reasons as well, this can diminish the dates’ value.
Two Israeli machinery manufacturers, Com-N-Sense and Lugo Engineering, jointly developed an automated inspection and sorting system for dates that optically inspects each individual fruit and sorts it by category.
First challenge even before inspection: feeding into the system
“Dates aren’t easy to work with,” says Lirian Shahar, Managing Director of Com-N-Sense. “In comparison to other fruits, they’re small, and yet there are significant differences in size among them. On top of that, they’re sticky, and after harvesting, they often get stuck together in clumps.” Before they can be inspected, they have to be fed into the system in a manageable way.
To separate sticky dates from each other, Com-N-Sense and Lugo developed an innovative process. Loose fruit is fed onto a vibrating feeder whose surface is covered with V-shaped grooves. The vibration separates the individual dates from each other and they line up behind each other in the grooves.
From there, the dates are fed onto a conveyor system consisting of two parallel polyester cords; the dates are placed between them, preventing the dates from moving during transit and the inspection process. Four parallel conveyor systems move the dates to the inspection station at a speed of 6 dates per second, allowing the system to inspect 24 dates per second.
Imaging system captures 1400 pieces per minute
Each conveyor moves through an inspection station consisting of a triggering sensor, a digital camera, and lighting. Four Prosilica GC1290C color cameras from Allied Vision Technologies are positioned over the four conveyor systems. Each camera is linked to a dome light with a 10 cm diameter, providing even, white illumination of the dates. This is especially important because the examination is based upon color analysis. The camera and lighting are triggered via photo sensor for every passing piece of fruit.
The AVT Prosilica GC1290C is an especially compact and robust digital camera with a GigE Vision Interface. It is equipped with a high-sensitivity color sensor from Sony (ICX445) utilizing Sony EXview HAD CCD technology with outstanding image quality. The Prosilica GC1290C delivers up to 32 images per second at full resolution (1.2 megapixels).
The captured images are transferred via the cameras’ Gigabit Ethernet interface (GigE Vision) to the host computers, where they are analyzed using proprietary imaging
software that Com-N-Sense programmed especially for this application, using elements of OpenCV.
The date is first distinguished from the background,” Shahar explained. “The enormous color contrast between the bright green polyester cords and the reddish-brown dates achieves this.” RGB data is then converted into HSI color-space to conduct color analysis. Blisters on the outer surface appear in a lighter color. In this manner, quantity, size and form of the blisters can be determined.
Beyond a certain threshold, dates that exhibit too many blisters are sorted out using compressed air. Selection criteria can be individually defined according to customer quality demands, and different categories can be set, such as product for processing, for retail sale, and so forth.
New functionality: moisture measurement
Com-N-Sense und Lugo Engineering’s system is already in successful use in Israel and is being further refined by its inventors. Lirian Shahar intends to provide one particular new functionality: the detection of moisture content within fruit. To achieve this, each individual date will need to be weighed while being measured: “Using image data, we can estimate the dates’ dimensions and calculate their volume based upon that. From the relationship of volume to weight, we can extrapolate moisture content — how juicy the individual date is. This is an important quality characteristic for our customers that they will be able to check with coming generations of this system.”
Com-N-Sense and Lugo Engineering were advised and serviced locally by OpteamX, Allied Vision Technologies‘ sales partner in Israel.
Machine vision makes automated fruit and vegetable sorting possible
Products of nature, such as apples and tomatoes, aren’t standardized: each one differs from the next in form, ripeness, optical qualities, weight, bruising, and so forth. The more exactly these quality characteristics can be measured, the more optimally the harvest can be marketed. From the gourmet market to the discount retailer and on to the processing industry, quality demands are different and prices are correspondingly high or low. For that reason, it’s in the producers’ best economic interest to sort precisely, ensuring that fruits and vegetables are not sold below their value to the wrong customer.
For many years, Allied Vision Technologies has been working worldwide with leading suppliers of produce sorting equipment. Each individual piece is measured and checked in such systems and then sorted into the appropriate category. With high-performance digital cameras and imaging software, many characteristics can be inspected in a fraction of a second — size, form, color, presence of twigs and leaves, markings, blemishes, bruising, etc. — and assigned to the appropriate category. Using mirrors that have been properly positioned, a single camera can capture a 360˚ image of an apple, providing seamless inspection. With infrared cameras, even details not visible to the naked eye can be detected, such as water content or concentration within fruit, providing insight into the degree of ripeness and making bruises visible that would otherwise not be detected by a conventional camera or the naked eye performing a surface inspection.