Satake Smart Sensitivity offers new possibilities for optical sorting
by Nobuyoshi 'Nick' Ikeda, Assistant Manager, Technical Division, Satake Corporation, Japan
Electric Sorting Machine Company (ESM), a Michigan based company later acquired by Satake, developed the world's first optical sorter approximately 90 years ago. Since then, optical sorters have been used in various industries such as rice milling, flour milling, tree nuts processing, etc.
Optical sorter performance has improved year on year by adopting various new technologies available. Among the technological innovations contributing to the improvement, image processing technology is a feature that has tended to be overlooked. However, the effective utilisation of information from modern optics such as higher than 2K pixel high-resolution cameras, IR cameras, and full-color cameras, is one of the key features directly affecting sorting performance by helping determine both good product and defects - often referred to as accepts and rejects respectively.
The more information the optical sorter receives from the optics, the more the sorting parameters and criteria can be set, and as a result, the better the sorting performance it can achieve. On the other hand, it may create more confusion to the human operators to adjust or set the equipment due to multiple complex parameters.
Satake Smart Sensitivity (3S) developed by Satake in 2011, is an innovative solution to the above problems and offers not only hassle-free equipment adjustment and setting but also provides better sorting performance than optical sorters with traditional image-processing technology.
Normally, the adjustment of the optical sorter is done by setting a sensitivity to tell the equipment what it needs to separate. The adjustment process is like drawing a border between the information signal of the good products and the defective product, the accepts and rejects respectively.
If the information signals from the camera are one or two, the signal is mapped in 2D, therefore, the operator is able to determine the best position of the sensitivity needed to remove the defects easily. However, in the current mainstream optical sorter, equipped with a full-color camera, the image obtained from the camera is represented by a combination of three types of information signals, red, green, and blue.
The operator has to find the most effective angle of the border in the 3D space in order to determine the best position for the sensitivity parameters. In other words, the operator needs to rotate the 3D space in which the signal group of accepts and rejects are mapped in all directions. This is an extremely difficult task compared to the traditional image processing of optical sorters in 2D space.
Satake Smart Sensitivity (3S) has a function to automatically locate the most effective "angles' for analysis. Users just simply select a few pieces of both good and defect grains and let the optical sorter 'learn' those images, and then the 3S can quickly find the most effective angles to determine sensitivities automatically.
With this function, the difficulty of setting the sensitivity is dramatically reduced and it becomes possible to optimise the performance of the optical sorter easily. In addition, by adopting the concept of rotation of the angle of this space, it becomes possible to distinguish the colour difference which cannot be distinguished by the 2D space, and the sorting performance is improved as a result.
For example, when two types of signals are expressed in 2D space, the clusters of accepts and rejects are partially overlapped. The operator will experience difficulties determining the position of the sensitivities. If the sensitivity is set at position A, the rejects will not be missed, but the probability of false detection of the accept as a reject will increase, resulting in an increase in the amount discharged as a defective product of the optical sorter, or a decrease in the product yield.
Conversely, if the sensitivity is set at position B, the accepts will not be incorrectly detected as rejects, but some rejects will get into the accepts.
Satake Smart Sensitivity can maximise the product yield without missing defects because it automatically finds the most suitable angles for determining the sensitivity for distinguishing between accepts and rejects in the 3D space.
One of the features of the latest optical sorters is that operators can set multiple sensitivities for multiple defects such as discoloured products or foreign materials. However, prior to these new technologies, there were no means to know the level to which each sensitivity was detected as a defect.
In other words, operators would not know how much the particular sensitivity is removing the corresponding defects. Therefore, the operator had to make a trial-and-error adjustment while looking at the output samples in order to find the degree to which sensitivity adjustment began affecting the sorting performance.
With Satake Smart Sensitivity, the operator is able to set up six different types of sensitivity based on the defect, such as discoloured products, Fusarium-contaminated grains, stones, etc. All sensitivities have the function of displaying the level of detection of the defect per material sample.
Defect level is expressed as a numerical value, indicating the frequency of the set sensitivity detecting the defects and it is visually displayed as a meter (A) and a trend graph (B) for each sensitivity. This feature allows the user to know in real time how sensitivity is affecting the sorting, so that the operator can intuitively and quickly optimise sorting performance.
Satake Smart Sensitivity, installed in optical sorters such as Satake's RGBR series and REZX series, provides an essential image processing technology for users to easily operate equipment, utilising multiple wavelength signals without requiring advanced skills and experience.
The technology allows users to set sensitivity by simply preparing samples of accepts and rejects as a "training dataset". Also, the indicator function displays the 'sorting level' information which is required for the operator to make sensitivity adjustments. These key functions will maximise the performance of the optical sorter and productivity of the users