NL2020014B1

NL2020014B1 - Potato length measurement

Info

Publication number: NL2020014B1
Application number: NL2020014A
Authority: NL
Inventors: Ferdinand Van De Merwe Mark
Original assignee: Lamb Weston/Meijer V O F
Priority date: 2017-12-04
Filing date: 2017-12-04
Publication date: 2019-06-11

Abstract

The present invention relates to a method and system for sorting an incoming stream of raw potatoes. The present invention proposes to set a threshold used during a second optical sorting step in dependence of an average potato length that is measured upstream of a first mechanical sorting step.

Description

Potato length measurement

The present invention relates to a method and system for sorting an incoming stream of raw potatoes. In most factories, the incoming stream of raw potatoes is sorted into one or more separate sub-streams, wherein each sub-stream holds potatoes for different products. For example, at the potato processing factory, one or more batches of raw potatoes may be deposited from the trailer used for transporting the potatoes onto a conveying system, thereby resulting in an incoming stream of potatoes. This incoming stream may be split into different sub-streams, which each hold potatoes for the production of different products or for the production of products for different clients. As an example, a primary stream may hold potatoes for the production of French fries, and a secondary stream may hold potatoes for the production of potato wedges. Typically, the length requirements for these sub-streams are different. A sorting process is therefore performed to split the incoming stream of raw potatoes into the various required sub-streams.

Within the context of the present invention, the shape of the potato is assumed to be an ellipsoid, wherein the length of the potato corresponds to the length of the largest principal axis of the ellipsoid.

In a known process for sorting the incoming stream of potatoes, the average potato length for a given potato batch is known prior to being subjected to the sorting process in the factory. For example, the average length of the potatoes is determined by the potato grower or is determined in the factory itself, for instance by manually measuring the potato length. By knowing the average length prior to processing of the batch, settings of the sorting process can be optimized.

To split the incoming stream of potatoes into a plurality of sub-streams, it is known to use two consecutive sorting processes. The most upstream sorting process comprises a mechanical sorting step wherein the stream of potatoes is transported along a vertical stack of sieves, wherein each sieve has a different size of the sieve openings. The potatoes are first subjected to the sieve having the largest opening. Potatoes that are smaller than the sieve opening will fall through the first sieve onto the next sieve in the stack having a smaller opening, and potatoes that are larger will be collected by a separate conveying unit, such as a conveying belt.

Typically, each sieve is associated with a separate conveying unit. It therefore becomes possible to sort the incoming potato stream into a plurality of different sub-streams based on the size of the potatoes.

Most potato-based products have well-defined size requirements. For example, a bag of pre-baked French fries should have a particular average size of the fries that falls within a given range, and the size of the French fries should have a given distribution. As an example, the average size should be 85 mm, and less than 2.5% of the fries should have a size smaller than 50 mm.

Although the mechanical sieve is capable of handling high throughputs, it is unable to sort the stream of potatoes to a sufficient degree. Therefore, an additional sorting step is required. This additional sorting step typically comprises an optical sorting step, wherein each individual potato is imaged and sorted. For example, an optical camera may be used to record each potato individually. Known image techniques may then be used to accurately determine the potato length. A drawback of the known optical sorting step is that it is unable to handle high throughputs. To solve the mismatch between the throughput of the incoming stream of raw' potatoes and the throughput that can be handled by the optical sorting step, only a relatively small portion of the incoming stream will be subjected to optical sorting.

Next, an example will be discussed wherein the incoming stream of raw potatoes will be sorted into a primary stream holding potatoes for the production of fries, and a secondary stream that is used for other products or for blending purposes in which potatoes of different lengths are combined.

In this example, the mechanical sorting step will result in six different sub-streams. The first and second sub-streams hold the largest potatoes. These sub-streams originate from the two sieves having the largest openings and are combined into the primary stream. A fifth and sixth substream hold potatoes that are too small for the production of French fries. These sub-streams originate from the two sieves having the smallest openings and are combined into the secondary stream. A third and fourth sub-stream hold potatoes that may or may not be suitable for the production of French fries. These sub-streams originate from the sieves having intermediate openings. These sub-streams are combined into a temporary stream which is subjected to the optical sorting step. Due to the mechanical sorting, the throughput of the temporary stream is considerably lower than the throughput of the original incoming stream. The temporary stream may therefore be subjected to the optical sorting step without causing throughput problems.

During the optical sorting step, images are acquired for each potato individually. Based on a length measurement, it is decided, for each potato individually, whether that potato should be directed to the primary or secondary stream.

The optical sorting step is characterized by one or more parameters, the most important parameter being the length threshold used for deciding whether the potato should be admitted to the primary stream (size > threshold) or to the secondary stream (size < threshold). In the know'll sorting process, the length threshold is set in dependence of the average potato length of the current batch of raw' potatoes. For example, if the grower responsible for a batch of raw potatoes has determined that the average length is relatively high, a lower threshold may be used for the optical sorting step as there will be a relatively large first and second sub-stream allowing shorter potatoes to be added to the primary stream to arrive at the desired average length.

The applicant has found that the efficiency of this known process for the sorting of potatoes is often too low. For example, the amount of potatoes in the secondary stream may be too high, resulting in loss of efficiency for the primary stream.

To solve this problem, the invention proposes a method for sorting an incoming stream of raw potatoes as defined in claim 1.

According to the invention, an average potato length is repeatedly measured for a portion of the potatoes in the incoming stream and upstream of the mechanical sorting step. Furthermore, the threshold for the optical sorting step, referred to as the third threshold, is set, preferably automatically, in dependence of this measured average potato length. Hence, instead of setting the third threshold in dependence of an average potato length that applies to the batch as a whole, it now becomes possible to account for variation in potato length during the processing of a batch. This will considerably improve the efficiency of the sorting process. For example, the first potatoes that are processed from a batch will display a relatively high length when compared to the remaining potatoes of such batch. Accordingly, a larger fraction of smaller potatoes may be added to the primary stream without causing the average length in the primary stream to drop below the desired value. In other words, when a higher average potato length is measured, it may be possible to set the third threshold to a lower value, and vice versa.

The average potato length at a given time l can be determined using potato length measurements performed in the time interval between t-dt and t, with dt corresponding to a predetermined time interval. For example, an average potato length for time 1=19:00:00, may correspond to measurements that were performed during the time interval between t=18:55:00 and t= 19:00:00. Alternatively, the average potato length can be determined using a predetermined number of last potato length measurements. In such a case, the average potato length for a time 1=19:00:00 may be determined using the last 100 measurements of the potato length.

According to the invention, the average potato length is repeatedly measured and used for setting the third threshold. The process of setting the third threshold may be a substantial continuous process, meaning that every change in the average potato length, regardless the time interval in which such change occurs, is reflected in a different setting of the third threshold. Alternatively, the third threshold is set at fixed time intervals, such as every five minutes.

The method may further comprise an optical measurement step for optically measuring the average potato length upstream of the mechanical sorting step. This step may comprise performing laser triangulation to determine a height profile for a plurality of potatoes in the incoming stream, and determining the average potato length based on the determined height profiles. For example, a plurality of rectangular boxes may be fitted to the determined height profiles to identify a plurality of potatoes. A length of each identified potato may then be determined based on the size of the rectangular box for that potato, and the average potato length may be determined by averaging the determined lengths. The averaging may be performed over time or over a given number of potatoes.

The optical measurement step may further comprise comparing each height profile to a predefined reference and determining whether that height profile should be included in the determination of the average potato length based on said comparison. For example, a potato that lies with its principal axis parallel to the conveyor belt on which it is transported, will, provided that it has a normal outer appearance, have a given shape, e.g. ellipsoid, and a certain size range. When the height profile shows deviations from that shape and/or size range, for example because the potato has an unusual shape or size or is positioned with its largest principal axis perpendicular to the conveyor belt, it may be decided to disregard the potato length measurement for that potato.

Performing laser triangulalion to determine a height profile may comprises scanning, with a beam of laser light, the incoming stream of raw potatoes using a line scan in a direction perpendicular to a transport direction of the incoming stream. For example, a laser source emitting a single beam of laser light may be moved, e.g. rotated, such that a line is formed on the conveyor belt and potatoes underneath the laser that is substantially perpendicular to the transport direction of the potatoes. The reflections of this line are captured by an optical detector, such an optical camera. A height position can be associated with each point at which the laser light was reflected based on, or in dependence of, the angle under which the reflected light reaches the optical detector.

Height offsets relative to a reference level may be determined using the line scan. The line scanning may be repeatedly performed while the incoming stream of potatoes moves in the transport direction. The thus obtained line scans may be collected. If the entire potato has been scanned using the laser, a height profile may be determined using the collected line scans.

Typically, the optical sorting step is performed some time after the average potato length has been determined. It may therefore be advantageous to set the third threshold taking into account a delay between the determination of the average potato length for a given portion of potatoes and the optical sorting of those potatoes.

According to a further aspect, the present invention provides a system for sorting an incoming stream of raw potatoes. This system comprises a conveying system for conveying the incoming stream of raw potatoes, a mechanical sorting unit for splitting the incoming stream of raw potatoes into a plurality of separate sub-streams based on the size of the potatoes, each substream being transported by the conveying system. The system further comprises a combining unit for combining one or more of the sub-streams into a primary stream representing a stream of potatoes having a length or size that exceeds a first threshold, and for combining one or more of the sub-streams into a secondary stream, representing a stream of potatoes having a length or size that is lower than a second threshold. The system also comprises an optical sorting unit for subjecting the sub-streams that have not been combined into the primary or secondary streams to an optical sorting step in which step a length of each individual potato is determined, the optical sorting unit being further configured to feed each optically measured potato to the primary stream or to the secondary stream based on a comparison between the determined length and a third threshold.

The system according to the invention is characterized in that it further comprises a length measurement unit arranged upstream of the mechanical sorting unit, which length measurement unit is configured for repeatedly measuring an average potato length for a portion of the potatoes in the incoming stream, and a control unit for setting, preferably automatically, the third threshold in dependence of the measured average potato length.

The length measurement unit may comprises a laser for emitting a beam of laser light that scans in a direction perpendicular to a transport direction of the incoming stream, and an optical detector, such as an optical camera, for detecting laser light that is reflected from the potatoes. A processing unit arranged in the length measurement unit may be used for controlling the laser and for collecting and processing signals from the optical detector, wherein the processing unit is configured to perform the steps as defined above in conjunction with laser triangulation.

The skilled person will readily understand that means of producing a focused beam of light other than a laser may equally be used.

Next, the invention will be discussed in more detail by referring to the appended drawings, wherein:

Figure 1 illustrates an embodiment of a system for sorting an incoming stream of raw potatoes in accordance with the present invention;

Figures 2A and 2B illustrate the concept of laser triangulation used in the system of figure 1;

Figure 3 illustrates examples of height profiles determined using the laser triangulation technique of figure 2; and

Figure 4 illustrates a method for sorting an incoming stream of raw potatoes in accordance with the present invention.

Figure 1 illustrates an embodiment of a system 1 for sorting an incoming stream of raw potatoes in accordance with the present invention. It comprises a conveying system 2 that may comprise a plurality of connected conveyor belts for guiding the various streams of potatoes. The system further comprise a mechanical sorting unit 3 that receives, via conveying system 2, an incoming stream of raw potatoes. This incoming stream may have already been subjected to various stages of processing, such as cleaning processes, but has not or has not completely been subjected to a sorting based on the size of the potatoes. Such sorting is important if a final product is desired in which the various potatoes or potato-parts need to have a predefined size distribution.

To this end, mechanical sorting unit 3 comprises a plurality of vertically stacked sieves, each sieve having a different size of the sieve opening, and the size of the opening decreasing when moving downwards in the stack. As an example, mechanical sorting unit 3 comprises a stack of 6 sieves, wherein the first sieve has a sieve opening of 15cm, the second sieve 12.5cm, the third sieve 10cm, the fourth sieve 7.5cm, the fifth sieve 5cm, and the sixth sieve 2.5cm. When guiding the incoming potatoes over the first sieve, potatoes having a size smaller than 15cm will fall through the sieve openings onto the second sieve. Potatoes that have a size larger than 15cm will not fall through the sieve opening but will be collected at the end of the first sieve by a conveyor belt. To increase the sieving operation, the sieves may perform a shaking motion.

In the exemplary mechanical sorting unit 3 described above, each sieve is connected to a respective conveyor belt. More in particular, mechanical sorting unit 3 outputs six different sub-streams of potatoes, namely a first sub-stream with potatoes having a size s >15cm, a second substream with 12.5cm <s<15 cm, a third sub-stream with 10cm<s<12.5cm, a fourth sub-stream with 7.5cm<s<10cm, a fifth sub-stream with 5cm<s<7.5cm and a sixth sub-stream with s <5cm. Next, combining unit 4 combines some of the sub-streams into a primary stream and a secondary stream. As an example, the primary stream may comprise the first and second sub-streams and the secondary steam may comprise the fifth and sixth sub-streams. The primary stream may correspond to potatoes that are all to be used for the intended product whereas the secondary stream may correspond to potatoes that are not to be used for the intended product. The third and fourth sub-streams are not used by combining unit 4. Instead, these sub-streams are guided, via conveying system 2, to optical sorting unit 5. This unit optically characterizes each potato by determining its size. This size is compared to a threshold. If the size is larger than the threshold, the potato is added to the primary stream via conveying system 2. If the size is smaller than the threshold, the potato is added to the secondary stream via conveying system 2. The present invention does not exclude that, if defects are found in the potato by means of optical characterization, for example abnormal pits or coloration, the respective potato is guided to a waste buffer or bin.

Optical sorting unit 5 may comprise a plurality of elongated rollers that are arranged in the transport direction of the potatoes. Due to rotation of the rollers, the potatoes are forced to assume a position with their longest principal axis in the direction of transport. Optical sorting unit 5 further comprises one or more optical camera to record images of the re-positioned potatoes. As the orientation of the potatoes corresponds to the desired orientation, a potato length for each potato can be accurately determined. However, inter alia due to the re-orientation process of the potatoes, the throughput of optical sorting unit 5 is relatively low.

System 1 further comprises a length measurement unit 6 that is arranged upstream of mechanical sorting unit 3. This unit is explained in more detail in figures 2A and 2B. Unit 6 comprises a source 61 of focused light, such as a laser, which emits a beam of light onto potatoes 100 that are passing on a conveyor belt 21 of conveying system 2 underneath unit 6. Source 61 is configured or controlled to produce a line of light on potatoes 100 and conveyor belt 21 as indicated by dotted line 61 A.

The light reflected from potatoes 100 is captured by an optical detector 62, preferably in the form of an optical camera. As shown in figure 2A, the angle under which light falls onto detector 62 is representative for the height position at which the light beam was reflected. In figure 2A, the beam indicated by the dotted line is reflected at a lower position relative to conveyor belt 21 than the beam indicated by the solid line. If detector 62 comprises a photosensitive surface, such as a photosensitive chip, both beams will be imaged at a different position on the surface. A height position for a given position along line 61A is repeatedly determined. As conveyor belt 21 is moving, the determined height position corresponds to a different location of a potato. In this manner, potatoes 100 can be fully scanned from a top side thereof. A processing unit 64, which may be arranged outside of a housing 63 in which detector 62 and source 61 are arranged, is used to process the data from detector 62. More in particular, processing unit 64 outputs a height profile 8 for potatoes 100 as illustrated in figure 3. A height profile indicates in a two-dimensional plane the various height positions, which may be expressed as a numerical value representing the distance between the position at which the corresponding light beam was reflected and the position on the conveyor belt 21 directly underneath that reflection position. An example of this distance is indicated by arrow H in figure 2A.

Figure 3 illustrates that various height profiles 8-8’” can be measured. For example, height profiles 8 and 8”’ each indicate a potato that has its largest principal axis arranged in parallel with conveyor belt 21, whereas height profiles 8’ and 8’ ’ each indicate a potato that has its largest principal axis arranged perpendicular to conveyor belt 21.

Each height profile may comprise curves 81 that correspond to a given height. As such, curves 81 can be designated as iso-height curves.

Length measurement unit 6 is configured to determine the average length of potatoes 100. This determination may be performed on a continuous or intermittent basis. For example, the average length may be determined using the last N measurements. Alternatively, the average length may be determined using the measurements in the time interval t-dt and t, with t being the current time and dt a predetermined time interval. Instead of continuously performing a measurement, a measurement may be performed at predefined time intervals, e.g. every five minutes.

To determine the length of a potato 100, height profile 8 may be used. As indicated in figure 3, the length of a potato 100 corresponds to the longest principal axis as indicated by arrow' L. This length can be determined by fitting a rectangle 9 around the iso-height curves 81 that correspond to the same potato and by then determining the length of the longest side of fitted rectangle 9.

In most cases, it may not be efficient or accurate to use each height profile for determining a length. For example, using height profiles 8’ and 8” would result in a length that is too small as the potato is measured from the wrong direction. Hence, the determined height profiles may be compared to a reference, both in shape and size, to first determine whether a height profile is suitable for determining a length. Only suitable height profiles may then be used to improve the accuracy. Using a reference also enables potatoes having an abnormal shape to be excluded from the length measurement.

System 1 further comprises a control unit 7 to control at least optical sorting unit 5. This unit may also control the various other components in the system. Control unit 7 is configured for automatically setting the threshold used by optical sorting unit 5 in dependence of the average potato length as measured by length measurement unit 6. For example, the threshold may be lowered if a higher average potato length is measured. If a given average potato length is desired for the primary stream and if the average potato length is higher, there will be more potatoes in the primary sub-stream having a relatively large size. It may then be advantageous to combine these potatoes with a larger fraction of smaller potatoes from optical sorting unit 5. This can be achieved by lowering the threshold used by unit 5.

Within the context of the present invention, automatically setting the threshold refers to an embodiment wherein control unit 7 obtains the average potato length from length measurement unit 6 and provides this value to optical sorting unit 5, essentially without human intervention. It should be noted that the invention does not exclude embodiments wherein an operator uses the average potato length for manually setting the threshold used in optical sorting unit 5. Furthermore, setting the threshold may be delayed relative to the determination of the corresponding average potato length to ensure that the optical sorting unit uses the threshold that applies to the portion of potatoes that it is currently handling. This may be advantageous if a considerable delay exists between measuring a potato upstream of the mechanical sorting unit and optically charactering this same potato by optical sorting unit 5. This is even more advantageous if the average potato length changes relatively rapidly in time.

System 1 may further include one or more buffers 8 for temporarily storing potatoes corresponding to the primary stream, the secondary stream, or a waste stream.

Figure 4 illustrates a method for sorting an incoming stream of raw potatoes in accordance with the present invention. In a first step S1, an average potato length for a portion of potatoes in an incoming stream is repeatedly measured. Downstream of this measurement, the incoming stream of raw potatoes is split in a step S2 into a plurality of separate sub-streams based on the size of the potatoes in a mechanical sorting step. In a next step S3, one or more of the sub-streams are combined into a primary stream representing a stream of potatoes having a length or size that exceeds a first threshold, and one or more of the sub-streams are combined into a secondary stream, representing a stream of potatoes having a length or size that is lower than a second threshold. It may also be possible that some potatoes are discarded as a waste stream. In figure 4, the various streams are illustrated as various streams corresponding to a distribution 10 of the potatoes.

Next, in a step S4, the sub-streams that have not been combined into the primary or secondary streams are subjected to an optical sorting step, in which step a length of each individual potato is determined and, based on a comparison between the determined length and a third threshold th3, each optically measured potato is fed to the primary stream or to the secondary stream.

According to the invention, third threshold th3 is set in a step S5, preferably automatically, in dependence of the measured average potato length.

In the description above, the present invention has been explained using detailed embodiments thereof. The skilled person will understand that the invention is not limited to these embodiments. Various modifications can be made to these embodiments without departing from the scope of the invention, which is defined by the appended claims and their equivalents.

Claims

A method for sorting an incoming stream of raw potatoes, comprising: splitting (S2) the incoming stream of raw potatoes into a plurality of separate sub-streams based on the size of the potatoes in a mechanical sorting step; combining (S3) one or more of the sub-streams into a primary stream that represents a stream of potatoes that have a length or size that exceeds a first threshold; combining (S3) one or more of the sub-streams into a secondary stream representing a stream of potatoes having a length or size lower than a second threshold; subjecting (S4) the sub-streams that are not combined in the primary or secondary streams to an optical sorting step, in which step a length of each individual potato is determined and based on a comparison between the determined length and a third threshold , each optically measured potato is supplied to the primary stream or to the secondary stream. characterized by repeatedly measuring (S1) an average potato length for a portion of the potatoes in the incoming stream and upstream of the mechanical sorting step and setting (S5) the third threshold, preferably automatically, depending on the measured average potato length .

The method of claim 1, wherein the third threshold is set to a lower value when a higher average potato length is measured and vice versa.

A method according to claim 1 or 2, wherein the average potato length at a given time t is determined using the potato length measurements performed in a time interval between t-dt and t, wherein dt corresponds to a predetermined time interval; or wherein the average potato length at a given time t is determined using a predetermined number of last potato length measurements.

Method according to one of the preceding claims, further comprising an optical measuring step for optically measuring the average potato length upstream of the mechanical sorting step.

The method of claim 4, wherein the optical measuring step comprises performing laser triangulation to determine a height profile for a plurality of potatoes in the incoming stream, and determining determining the average potato length based on the determined height profiles.

The method of claim 5, wherein the optical measuring step further comprises fitting a plurality of rectangular sections to the determined height profiles to identify a plurality of potatoes; and determining a length of each identified potato based on the size of the rectangular box for that potato; and determining the average potato length by averaging the determined lengths.

Method according to any of claims 5-6, wherein the optical measuring step further comprises: comparing each height profile with a predetermined reference and determining whether that height profile should be included in the determination of the average potato length based on said comparison .

A method according to any of claims 5-7, wherein said performing laser triangulation to determine a height profile comprises: a) scanning, with a beam of laser light, the incoming stream of raw potatoes using a line scan in the direction perpendicular to a transport direction of the incoming stream; b) determining height shifts ("offsets") relative to a reference level using the line scan; c) repeatedly performing steps a) and b) as the incoming stream of potatoes moves in the conveying direction and collecting the line scans thus obtained; d) determining the elevation profiles using the collected line scans.

A method according to any one of the preceding claims, comprising setting the third threshold taking into account a delay between the determination of the average potato length for a given portion of potatoes and the optical sorting of those potatoes.

A system (1) for sorting an incoming stream of raw potatoes comprising: a transport system (2) for transporting the incoming stream of raw potatoes; a mechanical sorting unit (3) for splitting the incoming stream of raw potatoes into a plurality of separate sub-streams based on the size of the potatoes, each sub-stream being transported through the transport system; a combining unit (4) for: combining one or more of the sub-streams into a primary stream representing a stream of potatoes that have a length or size that exceeds a first threshold; and combining one or more of the sub-streams into a secondary stream representing a stream of potatoes that have a length or size lower than a second threshold; an optical sorting unit (5) for subjecting the sub-streams that are not combined in the primary or secondary streams to an optical sorting step in which a length of each individual potato is determined, the optical sorting unit being further configured to each optically provide measured potato to the primary stream or to the secondary stream based on a comparison between the determined length and a third threshold; characterized in that the system further comprises: a length measuring unit (6) arranged upstream of the mechanical sorting unit, which length measuring unit is configured to repeatedly measure an average potato length for a portion of the potatoes in the incoming stream; and a control unit (7) for automatically setting the third threshold in dependence on the measured average potato length.

The system of claim 10, wherein the length measuring unit comprises: a laser (61) for emitting a sample of laser light that scans in a direction perpendicular to the transport direction of the incoming stream, an optical detector (62), such as an optical camera , for detecting laser light reflected from the potatoes, and a processing unit (64) for controlling the laser and for collecting and processing signals from the optical detector, wherein the processing unit is configured to perform the steps as defined in a of claims 5-9.