Sorting it out: Optical sorting adds value

by Chrystal Shannon
Share This:

Optical sorting of rice adds value to a miller’s end-product, and can quickly offset installation costs

by Sarah Bee, Sortex Ltd., U.K.

Each year around 600 million tonnes of rice is grown throughout the world. Prior to export, at least 30 million tonnes of this global harvest, approximately 2x1015 grains, are inspected grain by grain with optical sorting machines.

Many of the typical defects in rice, including black spots caused by insect damage, bran streaks, pale green immature kernels and discoloured grains, can be identified by their color. By identifying and removing defective grains and foreign material, producers can provide a consistent, premium quality product at increased margins, which naturally strengthens their competitive positioning. In fact, it has now become a basic prerequisite for all optical sorting machines to identify and remove all gross contaminants, such as paddy, glass, stones and insects.

Modern consumers are demanding increased quality, and food processors are finding it necessary to protect themselves against increasingly litigious consumers. Compounding this, in recent years, tighter E.U. and American Food and Drug Administration standards on food quality have been implemented.

Food processors benefit from using automated systems for food sorting, since a machine can maintain greater levels of consistency than manual techniques and frequently offers reduced labor costs.

Sorted rice generally trades at prices between U.S.$250 and U.S.$550 per tonne, some 1% to 5% higher than unsorted rice. The optical sorting process adds so much value to the final product that the initial cost of purchasing and setting up a sorting machine can often be recouped in a matter of months.

Optical sorting of rice typically is carried out at the end of the milling process. Approximately 40% of the weight of the paddy is lost during the milling process. This means that the amount of rice that reaches the optical sorting stage is considerably less than the input capacity of the processing plant. Millers often process, and consequently sort, more than one type of rice.


Automated separation and sorting of food products began in the 1880s, when crude electrostatic methods were employed to remove chaff and other light materials from cereal grain. These early systems used photodiodes or photomultiplier tubes to discriminate between the overall color of the product and foreign bodies, such as stones and glass. These detectors are still used to find large-scale color var-iations, for example, to identify nuts that have not been shelled.

Nowadays, almost all color sorters use high-speed, solid state, Charge Coupled Device (CCD) cameras that can detect blemishes less than 1 millimeter in size.

In a bulk sorting system, dry products, such as rice, grains, seeds, coffee and nuts, are fed from a vibrating hopper onto a flat, or channelled, gravity chute. Ideally, the feeding method should separate the product into a uniform sheet, or monolayer. The product then passes into an optical inspection area, where a decision on whether to accept or reject each item is made. Defective objects are removed from the product stream with a controlled blast of air from a linear array of pneumatic ejector valves, positioned across the line of view.

In monochromatic optical sorting, defects and foreign material can be identified by examining at the color of the product. More specifically, it requires selectively comparing the magnitude of light that is reflected at certain wavelength bands. The sorting machine rejects dark objects reflecting less than the signal threshold set by the operator. Conversely, the machine can also be set to reject light objects from dark material.

Product is illuminated with visible light from fluorescent tubes that provide the required intensity and uniformity for accurate color sorting. Optical band-pass filters (colored glass) can be placed in front of the camera lenses so that only selected wavelengths (colors) reach the CCD detector.

For example, white rice reflects 50% of blue light (the 450 nanometer to 550nm wavelength range), while yellow rice reflects half that amount, allowing an easy separation of the two. This effect can be emphasized by using a blue band-pass filter and blue fluorescent tubes. Hence it is possible to consistently discriminate between two objects, provided their reflective reflectance differs by more than 20%.


The usual method for removing unwanted items from the main product stream is with a blast of compressed air from a high-speed solenoid or piezoelectric valve, connected to a strategically positioned nozzle.

Pneumatic ejector valves must have rapid action, reliability, long lifetime — a minimum of one billion cycles — and mechanical strength. The fastest ejector valve (a Sortex patented piezoelectric design) operates at a frequency of 1 kHz, firing a pulse of air for 1 to 3 miliseconds. Ejectors operate at input pressures between 200 to 550kPa (30 to 80 psi), depending on the size of the object to be removed.

Typically, the ejection point is located outside the optical inspection area, because the action of the air blast on a rejected object could cause dust particles and skin fragments to be blown around that could create false rejections. However, at the same time, it is advantageous to eject objects as soon as possible after the optical inspection point, due to unavoidable variations in the trajectory of each individual item.

Electronic circuits generate the appropriate time delay between the inspection and ejection points. Accurate timing is required to coincide the ejector air blast with that of the object to be ejected. This relies on the objects having constant velocity as they fall in front of the ejector nozzle. In practice, the tolerable variation in product velocity is about 10%. The trajectory of each particle also becomes harder to predict, the greater the distance between the viewing point and the ejection point. It can become a major design challenge to position the chute, optics and ejection system as close together as possible.

Food processing is usually a 24 hour-a-day, year-round operation. Operators cannot afford to regularly shut down a machine for even a few minutes to replace faulty ejectors. Under these circumstances, machine reliability and stability of operation are critical.


There is a frequently misconceived view that an optical sorting machine is guaranteed to remove all defects from a given batch of product. In reality, this is simply not achievable. Optical sorting will reduce the concentration of defective product, but it will never achieve 100% perfection.

All optical sorters are bound to remove some acceptable product and fail to remove some of the defective product.

There are a number of reasons for this limitation. Sometimes, the physical size or the color difference of the defect from the product may be too small for accurate detection. Occasionally the machine may detect a defect and remove the object, but the object re-enters the "accept" stream after it has been ejected as a consequence of a random collision. Ejector performance and minimal positional pitch of the ejectors in the array below the optical system can also become a limitation for accurate ejection.

A machine can be adjusted by its operator to optimize sorting performance.

Sensitivity is one of the principal parameters that the operator can change. Increasing the sensitivity will result in the machine rejecting more defective material. However, a greater proportion of good product will also be rejected as the sensitivity threshold approaches the average product color. There is normally a compromise point between achieving a high sorting efficiency and optimum yield (the ratio of good to bad material that is rejected). This compromise point is primarily achieved as a result of operator experience and training.

There are physical limits to the product throughput that a sorting machine can successfully achieve. If the product flow is increased above the upper limit, the "product sheet" flow will no longer be a single layer. Objects will overlap and sorting performance will deteriorate since many defects will be obscured and therefore, will not be detected by the optical system. Increasing the flow of product through the machine will also result in increased good product being lost, since overlapping and colliding products are difficult to eject efficiently.

Color sorting wheat

In wheat mills, a series of indent cylinders is used to separate brokens and to remove some contaminating weed and wild oat seeds. The oat seeds are large and are removed efficiently. However, the black weed seeds are smaller than the wheat grains, and the indent cylinder inevitably removes a lot of valuable broken wheat grains as well. It is then impractical to separate the broken wheat from the weed seeds, so this wheat is wasted.

In Korea, the idea of using color sorters instead of the traditional mechanical separators has been put into practice. At a Buhler wheat mill, the small indent cylinder was removed and replaced with a Sortex series 90000 color sorter, the same kind that is typically used around the world for rice sorting.

It was shown that the color sorter could eject all the black weed seeds while not removing any broken wheat. At the same time, it removed any remaining foreign material and unhulled grains. The result was a very clean product with minimal broken wastage, improving the yield of the mill by 0.5%.

Several other mills have now followed suit as the production and profit benefits were realized.