Foreign material identifier and removal confirmation device

Abstract

A method and apparatus for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, the method comprising: obtaining at least one image of suspected foreign material detected in a stream of good product containing some population of foreign material; performing image analysis on the at least one image of the suspected foreign material; and classifying, based on the image analysis, the suspected foreign material as true foreign material or non-foreign material; and a method and apparats for confirming the removal of a foreign object from a product flow during a sorting process, the method comprising: obtaining at least one image of at least one product removed from a product flow by a sorting system; comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object does not match the at least one image of at least one product removed from the product flow by the sorting system.

Description

TitleForeign Material Identifier and Removal Confirmation DeviceField of the Invention

[0001] The present invention relates to product sorting, and in particular to the identification and removal of foreign material during product sorting.Background to the Invention

[0002] In the field of food sorting and other bulk product sorting in general the removal of foreign material is a major concern. Foreign material can cause damage to product and equipment, even causing more foreign material in the product as it continues in the process line. So, removal of these foreign objects is imperative. Line operators and production companies see this as so important that they want specific alerts from the sorting equipment when foreign material is identified, however these alerts become a nuisance when there are too many false alerts or false positive signals. Depending on the product being processed and process conditions, process false positives can outnumber the real foreign material condition by hundreds to one. A device is required to inspect and qualify the signals indicating foreign material to be sure the alert is raised on real foreign material, and other signals should be ignored. This device to do a secondary inspection of the signals can take longer to analyze the signal in more detail and use other more time-consuming analysis techniques that would be possible where a sorting machine must make multiple decisions as product is travelling to an ejection point.

[0003] A second related problem is the verification that the foreign material was in fact removed from the product stream. This produces a problem in bulk sorting that the product is randomly placed on the infeed transport system, and after flowing through the sort / ej ection zone the position of any item will be changed with respect to the flow and other product. If an object is ejected from the flow its position in the ejection stream will be randomized in that its emergence on an ejected product transport may be immediate, or may be delayed, depending on where it randomly landed on the transport and the speed of the transport system. The orientation of the object will not match the orientation when it was inspected and being a 3 -dimensional, object the visible aspect of the object may be very different from the aspect that was visible when originally inspected.Summary of the Invention

[0004] According to the present invention there is provided a method for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, the method comprising obtaining at least one image of suspected foreign material detected in a stream of good product containing some population of foreign material; performing image analysis on the at least one image of the suspected foreign material; and classifying, based on the image analysis, the suspected foreign material as true foreign material or non-foreign material.

[0005] The method may further comprise generating a signal indicating that the suspected foreign material has been classified as true foreign material, when the suspected foreign material is classified as true foreign material.

[0006] The method may further comprise generating a signal indicating that the suspected foreign material has been classified as non-foreign material, when the suspected foreign material is classified as non-foreign material.

[0007] Performing image analysis may comprise using a Machine Learning model trained on images of true foreign material and non-foreign material.

[0008] Performing image analysis may comprise performing at least one of an processing function to identify shape, an image processing function to identify size, an image processing function to identify colour, an image processing function to identify texture, and an image processing function to classify materials using Multivariate image analysis.

[0009] The at least one image of foreign material may be a multispectral or hyperspectral image with spectral bands within the range of 350nm to 1700nm.

[0010] The method may further comprise storing an image of any suspected foreign material classified as true foreign material.

[0011] The method may further comprise using the signal indicating that the suspected foreign material has been classified as true foreign material to actuate a secondary ejection device.

[0012] The method may further comprise providing an image of any suspected foreign material classified as true foreign material to at least one operator.

[0013] According to the present invention there is further provided an apparatus for identifying at least one image of foreign material from a plurality of images ofsuspected foreign material from a sorting system, the apparatus comprising: means for obtaining at least one image of suspected foreign material detected in a stream of good product containing some population of foreign material; means for performing image analysis on the at least one image of the suspected foreign material; and means for classifying, based on the image analysis, the suspected foreign material as true foreign material or non-foreign material.

[0014] The apparatus may further comprise means for generating a signal indicating that the suspected foreign material has been classified as true foreign material, when the suspected foreign material is classified as true foreign material.

[0015] The apparatus may further comprise means for generating a signal indicating that the suspected foreign material has been classified as non-foreign material, when the suspected foreign material is classified as non-foreign material.

[0016] The means for performing image analysis may comprise means for using a Machine Learning model trained on images of true foreign material and non-foreign material.

[0017] The means for performing image analysis may comprise means for performing at least one of an processing function to identify shape, an image processing function to identify size, an image processing function to identify colour, an image processing function to identify texture, and an image processing function to classify materials using Multivariate image analysis.

[0018] The at least one image of foreign material may be a multispectral or hyperspectral image with spectral bands within the range of 350nm to 1700nm.

[0019] The apparatus may further comprise means for storing an image of any suspected foreign material classified as true foreign material.

[0020] The apparatus may further comprise means for using the signal indicating that the suspected foreign material has been classified as true foreign material to actuate a secondary ejection device.

[0021] The apparatus may further comprise means for providing an image of any suspected foreign material classified as true foreign material to at least one operator.

[0022] According to the present invention there is further provided a method for confirming the removal of a foreign object from a product flow during a sorting process, the method comprising: obtaining at least one image of at least one product removed from a product flow by a sorting system; comparing the at least one image of the at leastone removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and determining a successful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product removed from the product flow by the sorting system. The least one image of a foreign object in the product flow identified for removal by the sorting system may be obtained from the sorting system. The at least one image of at least one product removed from a product flow by a sorting system may be obtained using a camera.

[0023] Obtaining at least one image of at least one product removed from a product flow by a sorting system may comprise recording the at least one image of at least one product using a camera positioned along a reject path exiting the sorting system.

[0024] Obtaining at least one image of at least one product removed from a product flow by a sorting system may comprise obtaining a plurality of images of a plurality of objects removed from the product flow.

[0025] The method may further comprise signalling the determination of the successful removal.

[0026] The method may further comprise obtaining at least one image of at least one product retained in the product flow by the sorting system.

[0027] The method may further comprise: comparing the at least one image of the at least one retained product with at the least one image of a foreign object in the product flow identified for removal by the sorting system; and determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product retained in the product flow by the sorting system.

[0028] The method may further comprise signalling the determination of the unsuccessful removal.

[0029] Obtaining at least one image of at least one product retained in the product flow by a sorting system may comprise recording the at least one image of at least one product using a camera positioned along an accept path exiting the sorting system.

[0030] According to the present invention there is provided a method for confirming the removal of a foreign object from a product flow during a sorting process, the method comprising: obtaining at least one image of at least one product removed from a product flow by a sorting system; comparing the at least one image of the at least one removedproduct with at least one image of a foreign object in the product flow identified for removal by the sorting system; and determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object does not match the at least one image of at least one product removed from the product flow by the sorting system.

[0031] Determining an unsuccessful removal of the foreign object from the product flow may comprise determining the unsuccessful removal of the foreign object from the product flow at the end of a time period.

[0032] The time period may start at a time of identification by the sorting system of the foreign object in the product flow. The time period may depend on the speed of flow of product through the sorting system, and / or based on when the product is expected to have passed through the sorting system.

[0033] Any of the method steps and apparatus features described herein may be used alone or in combination, in any order, in accordance with any aspect of the present invention.

[0034] According to the present invention there is provided an apparatus for confirming the removal of a foreign object from a product flow during a sorting process, the apparatus comprising: means for obtaining at least one image of at least one product removed from a product flow by a sorting system; means for comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and means for determining a successful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product removed from the product flow by the sorting system.

[0035] The means for obtaining at least one image of at least one product removed from a product flow by a sorting system may comprise means for recording the at least one image of at least one product using a camera positioned along a reject path exiting the sorting system.

[0036] The means for obtaining at least one image of at least one product removed from a product flow by a sorting system may comprise means for obtaining a plurality of images of a plurality of objects removed from the product flow.

[0037] The apparatus may further comprise means for signalling the determination of the successful removal.

[0038] The apparatus may further comprise means for obtaining at least one image of at least one product retained in the product flow by the sorting system.

[0039] The apparatus may further comprise: means for comparing the at least one image of the at least one retained product with at the least one image of a foreign object in the product flow identified for removal by the sorting system; and means for determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product retained in the product flow by the sorting system.

[0040] The apparatus may further comprise means for signalling the determination of the unsuccessful removal.

[0041] The means for obtaining at least one image of at least one product retained in the product flow by a sorting system may comprise means for recording the at least one image of at least one product using a camera positioned along an accept path exiting the sorting system.

[0042] According to the present invention there is further provided an apparatus for confirming the removal of a foreign object from a product flow during a sorting process, the apparatus comprising: means for obtaining at least one image of at least one product removed from a product flow by a sorting system; means for comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and means for determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object does not match the at least one image of at least one product removed from the product flow by the sorting system.

[0043] The means for determining an unsuccessful removal of the foreign object from the product flow may comprise means for determining the unsuccessful removal of the foreign object from the product flow at the end of a time period.

[0044] The time period may start at a time of identification by the sorting system of the foreign object in the product flow.

[0045] According to the present invention there is provided a computer readable medium comprising instructions thereon which when executed by a processor cause the processor to carry out any of the aforementioned methods.

[0046] Given there is a reliable signal from the post sorting foreign material qualifier a secondary similar foreign material qualifier would inspect the ejected product stream tomatch the signal as sorted with the signal of items being transported in from the ejection system. If a match is detected than the sorting system can positively confirm ejection of the foreign material. A similar secondary qualifier system can be inspecting the product on the accept side of the sorter ejection system, in case the foreign material was missed by the ejection system, and an alert can be raised that there is still foreign material in the acceptable product. If the foreign material is not identified in either stream it is possible that the foreign material was ejected and, in some way escaped from the system (bounce / ricochet) or was damaged beyond recognition by the ejection device or was occluded by other product in either ejection or accept stream.

[0047] With a bulk flow of products through an inspection zone, an inspection method / device may be used to detect foreign material with or on the bulk flow of product passing through the inspection zone. An ejection method / device may be activated by a signal from the inspection method / device when a foreign material is detected with a high degree of certainty, which ejects the detected material to a transport means to a second inspection method / device. The second inspection device may detect objects ejected and match them to the objects detected by the first inspection method / device. A method to communicate that the foreign material was ejected and confirmed to be detected by the second inspection method / device may also be provided.

[0048] An optional third inspection method / device may detect the bulk flow of products after the first inspection and ejection methods / devices and attempt to match any detected foreign material in case it was missed or the ejection method / device failed to eject from the bulk flow.

[0049] Inspection methods can be camera systems which can be monochrome, colour, colour with infrared channels, hyperspectral cameras, and or multispectral cameras. Illumination for the inspection systems can be monochrome, broad spectrum, multiple narrow band, etc. Illumination sources may be constant on or pulsed.

[0050] Inspection methods may refer to sensors using electromagnetic radiation (light) either reflected from or transmitted through the product being inspected, the reflected or transmitted radiation being focused on a sensor to convert to signals to be digitized and forming a digital image of the product being inspected.

[0051] Inspection methods can combine multiple views of the product flow to inspect the top and bottom surfaces of the product. Inspection can include multiple views of the product flow from different angles to inspect multiple aspects of the product

[0052] Depending on the foreign material and product being inspected the Electro Magnetic (EM) radiation or light, used to illuminate the product for the camera or detector, can be in the UV Visible and Infra-red regions of the spectrum, ranging from 380nm to 1700nm. In the case of inspection by transmission of EM radiation through product the spectrum is extended to the X-Ray bands and can include dual energy X- Ray imaging techniques.

[0053] Foreign material detection can be produced by a number of technologies using image processing and analysis of images output by the inspection sensors. The second inspection and the optional third inspection unit(s) may match the sensing technologies on the primary sensor sufficiently so that images from the subsequent inspection devices can be meaningfully compared to the primary inspection system.

[0054] Classification of objects in images is possible using a neural net classifier trained with images from the primary inspection device with sufficient spectral channels and resolution to identify the foreign material with respect to the good product and other product present on the production line.

[0055] Cloud based repositories for image data to create classifier models are most suited for this invention because images can be collected from multiple connected locations in different geographies, the more variety the models are trained with, the more robust the classifier. Similarly the classifier model learns from images of good product and what is not foreign material, collecting the images of non-foreign material from a wide geographical base and from many inspection machines reduces the chances of over learning and bias to particular varieties or local conditions.

[0056] There is further provided, in accordance with one aspect of the present invention, an apparatus for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system comprising: a sorting system for detecting foreign material from a stream of good product containing some population of foreign material; a means of communicating images of the detected foreign material; a computer with memory and processing capability suitable of image analysis and communication means suitable to receive said communicated images,

[0057] Wherein said computer also comprises communications to signal the true identification for foreign material or a false identification based on the said image analysis.

[0058] The said image analysis process may comprise an inference engine and Machine Learning model trained on images of true foreign material and the non-foreign material.

[0059] The image analysis process may comprise image processing functions to identify shape, image processing functions to identify size, image processing functions to identify colour, image processing functions to identify texture, image processing functions to classify materials using Multivariate image analysis.

[0060] The images communicated of foreign material may be multispectral or hyperspectral images with spectral bands within the range 350nm to 1700nm.

[0061] The image of the said true identification may be transmitted to a database and stored.

[0062] The communications of a said true identification may comprise a signal to actuate a secondary ejection device.

[0063] The communications and the said images representing the true identification may be notified to operators.

[0064] The computer and processing capability may comprise at least one GPU Graphics Processing Unit or APU Accelerated Processing Unit for implementing the neural network associated with the Machine Learning Model.

[0065] The GPU or APU may be incorporated in the Central Processing Unit (CPU).

[0066] There is further provided, in accordance with one aspect of the present invention, an apparatus for confirming the output of a sorting system(l 1), comprising a means of getting images from said sorting system of objects being ejected (12), a communications means for getting images from said sorting system of objects being accepted (13), at least one camera and illumination to record images of products ejected by the sorting system (14), optionally at least one camera and illumination to record images of products accepted by the sorting system (15), an image processing computer (16) that matches the image of the objects intended to be ejected with the images of objects recoded by said camera (14) and confirms it is not found in images recorded by said camera (15), a communications method to signal the success or failure of the sorting ejection.Brief Description of the Drawings

[0067] Embodiments of the invention will be described, by way of example only, with reference to the accompanying drawings in which:

[0068] Figure l is a representation of an apparatus for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, in accordance with one embodiment of the invention;

[0069] Figure 2 is a representation of an apparatus for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, in accordance with one embodiment of the invention;

[0070] Figure 3 is a representation of an apparatus for confirming the removal of a foreign object from a product flow during a sorting process, and / or for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system in accordance with one embodiment of the invention;

[0071] Figure 4 is a representation of an apparatus for confirming the removal of a foreign object from a product flow during a sorting process, and / or for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system in accordance with one embodiment of the invention;

[0072] Figure 5 is a representation of an apparatus for confirming the removal of a foreign object from a product flow during a sorting process, and / or for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system in accordance with one embodiment of the invention.Detailed Description of the Drawings

[0073] Identification of foreign material, also referred to herein as foreign objects or foreign material objects has been a requirement in the sorting processing industry for many years, however until recently the occurrence of too many false positives made this signal more of a nuisance than a valuable function. One aspect of the present invention sees the use of neural net inference technology for pattern recognition and image processing to learn and positively discriminate the false signals from actual foreign material signals. The positive signal and the same neural net technology may then be used to perform image analysis on the ejection and accept product streams to positively confirm the ejection of the foreign body, and if this cannot be confirmed to raise an alarm. The invention may further extend to compiling statistics from the foreign material and producing reports on the product quality and traceability reports on what was removed from the product and the like.

[0074] The present invention provides a device to positively qualify foreign material from normal good product and other signals that cause false positive foreign material signals to allow a reliable and robust foreign material alert signal to the process plant and external systems. This signal may then also be used to trigger other inspection and qualifier systems to look for and identify the same object as was ejected in the ejection transport system at some time later, to positively confirm capture of the foreign object in the ejection system. An optional secondary qualifier may be used to inspect the acceptable product to make sure the identified foreign material, or art thereof, was not missed by the ejection mechanism.

[0075] The invention may be used to report reliable foreign material identification to the plant, and to also signal confirmation of ejection, or an extra alert if the foreign object is still detected in the accept product stream. The value of this is in food safety and security, confirmation of sorter function, quality control and reporting of product quality, and audit trail of foreign material detection. Advantages include reducing product recall, and reducing maintenance due to damaged process machinery.

[0076] The present invention may be used in conjunction with known methods of foreign matter analysis and identifiers. Aspect of artificial intelligence (Al) and the cloud can be incorporated into the invention. Known foreign matter (FM) alerts may be coupled with images of the FM, which could be stored and retrieved from the cloud. This invention concentrates on confirmation and reporting of FM being ejected, and warning if the ejection could not be confirmed.

[0077] Figure 1 illustrates one embodiment of an apparatus for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, in accordance with one embodiment of the invention. An optical sorting system (1) is deployed in a product processing line to detect and remove unwanted foreign material from the incoming raw product (2). The output of the sorting system is sorted product (3) and ejected foreign material (4). The sorting system is designed and configured to transmit images of detected foreign material (7) which will be ejected. These images are communicated to the FM Analyzer (5) which comprises a computer, memory, storage, neural net acceleration processing devices like GPU or FPGA devices. The computer also comprises a number of communications interfaces, at least on communications device (8) to connect the computer to a repository (6) for images and inference models for the Neural Net, the figure indicates this repository iscloud based and connected through the internet, but other options are possible as illustrated in Fig 2. There are other options possible, and the location or connection of the repository is not material to the invention. The computer (5) also comprises communications interface and means for data and control (9) intended for the connection to automation and control systems in the product processing line. The computer (5) comprises a further communications interface and means to signal true detection of foreign material (10) this interface can communicate simple status messages and if required, images of the detected foreign material.

[0078] Figure 2. illustrates a similar system as that shown in Figure 1 except the repository for FM images and models for the neural network inference engines are stored in a local or remote computing and storage system, but not connected through the internet, for instance where an added level of security is required or the processing line is required to be self-contained. This configuration of the invention is restricted and, for instance, cannot as easily generate and deploy new inference models based on image data from multiple sites.

[0079] Figure 3 shows a system which extends the embodiment of Figure 2 and adds an inspection device (11) which produces images of the ejected product in a comparable format to the images (7) produced by the Optical Sorting System (1), in accordance with a further embodiment of the invention. These images of ejected objects (13) are compared to an image of the FM ejected by the sorter (14) sent from the FM Analyzer (5) by an image matching unit (12) comprising a computer, memory, storage, neural net acceleration processing devices. The computer also comprises communications interfaces for the receiving of the images and for communicating whether a match is found or not. The table in Fig 3 indicates the output of the device, the significance of a confirmed miss might be severe enough that an alarm would be raised or even that the production line is stopped. Not illustrated in figure 3 is the probable connection of the output signal (15) back to Neural Net FM Analyzer (5) so that the result produced by image matching (12) can be communicated to the processing line on communications interface and medium (9). Note it is possible that images of FM ejected (14) may contain all such images sent (7) or just the true FM images determined by the FM Analyzer.

[0080] Figure 4 builds directly on the embodiment illustrated in Figure 3 with the inclusion of an additional imaging device (16) on the accept output (3) of the OpticalSorting Machine (1) which produces images of all the objects accepted through the sorting machine, in accordance with a further embodiment of the present invention. These images are communicated to another image matching unit (18) of similar content to the image matching unit (12) described in Fig 3., it is possible that the function of both image matching units (13) and (18) could be performed within the one image processing system, processing the images from each source either in series or in parallel. The output (19) of this accept side image matching unit (18) can support the corresponding output (15) in the eject side. This signal can confirm correct ejection, confirm a missed ejection or in the case of neither secondary inspection unit detecting the Foreign Material, raise concerns about the containment of the product and the detectability of the foreign material. Communications (15) and (19) may be connected back to Neural Net FM Analyzer (5) so that the result produced by image matching (12) can be communicated to the processing line on communications interface and medium (9).

[0081] Figure 3 and 4 indicate the repository (6) for images and models is local or remote, this is just for illustration purposes and the repository (6) could also be cloud based and connected via the Internet. With reference to the accompanying drawings there is further provided an apparatus for identifying images of foreign material from a sequence of images of suspected foreign material from a sorting system comprising a sorting system (1) for detecting foreign material from a stream of good product containing some population of foreign material, a means of communicating (2) images of the detected foreign material, a computer (3) with memory (4) and processing capability (5) suitable of image analysis (6) and communication means (7) suitable to receive said communicated images (2), said computer (3) also comprises communications to signal (8) the true identification (9) for foreign material or a false identification (10) based on the said image analysis (6).

[0082] The image analysis process (6) may comprise an inference engine (11) and Machine Learning model (12) trained on images of true foreign material and the non- foreign material.

[0083] The image analysis process (6) may comprise image processing functions to identify shape (13), image processing functions to identify size (14), image processing functions to identify colour (15), image processing functions to identify texture (16), image processing functions to classify materials using Multivariate image analysis.

[0084] The images communicated of foreign material (2) could be multispectral or hyperspectral images with spectral bands within the range 350nm to 1700nm.

[0085] The image of the said true identification (9) may be transmitted to a database and stored.

[0086] The communications (8) of a said true identification (9) may comprise a signal to actuate a secondary ejection device.

[0087] The communications (8) and the said images (2) representing the true identification (9) may be notified to operators.

[0088] The computer (3) and processing capability (5) may comprise at least one GPU (17) Graphics Processing Unit or APU (18) Accelerated Processing Unit for implementing the neural network associated with the Machine Learning Model (12).

[0089] The GPU or APU may be incorporated in the Central Processing Unit (CPU).

[0090] There is further provided an apparatus for confirming the output of a sorting system(l 1), comprising a means of getting images from said sorting system of objects being ejected (12), a communications means for getting images from said sorting system of objects being accepted (13), at least one camera and illumination to record images of products ejected by the sorting system (14), optionally at least one camera and illumination to record images of products accepted by the sorting system (15), an image processing computer (16) that matches the image of the objects intended to be ejected with the images of objects recoded by said camera (14) and confirms it is not found in images recorded by said camera (15), a communications method to signal the success or failure of the sorting ejection.

[0091] The words “comprises / comprising” and the words “having / including” when used herein with reference to the present invention are used to specify the presence of stated features, integers, steps or components but do not preclude the presence or addition of one or more other features, integers, steps, components or groups thereof.

[0092] It is appreciated that certain features of the invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable sub-combination.

Claims

Claims1. A method for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, the method comprising: obtaining at least one image of suspected foreign material detected in a stream of good product containing some population of foreign material; performing image analysis on the at least one image of the suspected foreign material; and classifying, based on the image analysis, the suspected foreign material as true foreign material or non-foreign material.

2. The method of claim 1 further comprising generating a signal indicating that the suspected foreign material has been classified as true foreign material, when the suspected foreign material is classified as true foreign material.

3. The method of claim 1 further comprising generating a signal indicating that the suspected foreign material has been classified as non-foreign material, when the suspected foreign material is classified as non-foreign material.

4. The method of any preceding claim wherein performing image analysis comprises using a Machine Learning model trained on images of true foreign material and non-foreign material.

5. The method of any preceding claim wherein performing image analysis comprises performing at least one of an processing function to identify shape, an image processing function to identify size, an image processing function to identify colour, an image processing function to identify texture, and an image processing function to classify materials using Multivariate image analysis.

6. The method of any preceding claim wherein said at least one image of foreign material is a multispectral or hyperspectral image with spectral bands within the range of 350nm to 1700nm.

7. The method of any preceding claim further comprising storing an image of any suspected foreign material classified as true foreign material.

8. The method of any of claims 2-7 further comprising using the signal indicating that the suspected foreign material has been classified as true foreign material to actuate a secondary ejection device.

9. The method of any preceding claim further comprising providing an image of any suspected foreign material classified as true foreign material to at least one operator.

10. An apparatus for identifying at least one image of foreign material from a plurality of images of suspected foreign material from a sorting system, the apparatus comprising: means for obtaining at least one image of suspected foreign material detected in a stream of good product containing some population of foreign material; means for performing image analysis on the at least one image of the suspected foreign material; and means for classifying, based on the image analysis, the suspected foreign material as true foreign material or non-foreign material.

11. The apparatus of claim 10, further comprising means for generating a signal indicating that the suspected foreign material has been classified as true foreign material, when the suspected foreign material is classified as true foreign material.

12. The apparatus of claim 10, further comprising means for generating a signal indicating that the suspected foreign material has been classified as non-foreign material, when the suspected foreign material is classified as non-foreign material.

13. The apparatus of claim 10 or claim 11, wherein said means for performing image analysis comprises means for using a Machine Learning model trained on images of true foreign material and non-foreign material.

14. The apparatus of any of claims 10 to 13, wherein said means for performing image analysis comprises means for performing at least one of an processing function to identify shape, an image processing function to identify size, an image processing function to identify colour, an image processing function toidentify texture, and an image processing function to classify materials using Multivariate image analysis.

15. The apparatus of any of claims 10 to 14, wherein said at least one image of foreign material is a multispectral or hyperspectral image with spectral bands within the range of 350nm to 1700nm.

16. The apparatus of any of claims 10 to 15, further comprising means for storing an image of any suspected foreign material classified as true foreign material.

17. The apparatus of any of claims 11 to 16, further comprising means for using the signal indicating that the suspected foreign material has been classified as true foreign material to actuate a secondary ejection device.

18. The apparatus of any of claims 10 to 17, further comprising means for providing an image of any suspected foreign material classified as true foreign material to at least one operator.

19. A method for confirming the removal of a foreign object from a product flow during a sorting process, the method comprising: obtaining at least one image of at least one product removed from a product flow by a sorting system; comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and determining a successful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product removed from the product flow by the sorting system.

20. The method of claim 19, wherein obtaining at least one image of at least one product removed from a product flow by a sorting system comprises recording the at least one image of at least one product using a camera positioned along a reject path exiting the sorting system.

21. The method of any of claims 19 to 20, wherein obtaining at least one image of at least one product removed from a product flow by a sorting system comprises obtaining a plurality of images of a plurality of objects removed from the product flow.

22. The method of any of claims 19 to 21, further comprising signalling the determination of the successful removal.

23. The method of any of claims 19 to 22, further comprising obtaining at least one image of at least one product retained in the product flow by the sorting system.

24. The method of claim 23, further comprising: comparing the at least one image of the at least one retained product with at the least one image of a foreign object in the product flow identified for removal by the sorting system; and determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product retained in the product flow by the sorting system.

25. The method of claim 24, further comprising signalling the determination of the unsuccessful removal.

26. The method of any of claims 23 to 25, wherein obtaining at least one image of at least one product retained in the product flow by a sorting system comprises recording the at least one image of at least one product using a camera positioned along an accept path exiting the sorting system.

27. A method for confirming the removal of a foreign object from a product flow during a sorting process, the method comprising: obtaining at least one image of at least one product removed from a product flow by a sorting system; comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object does not match the at least one image of at least one product removed from the product flow by the sorting system.

28. The method of claim 27, wherein determining an unsuccessful removal of the foreign object from the product flow comprises determining the unsuccessful removal of the foreign object from the product flow at the end of a time period.

29. The method of claim 27 or claims 28, wherein the time period starts at a time of identification by the sorting system of the foreign object in the product flow.

30. An apparatus for confirming the removal of a foreign object from a product flow during a sorting process, the apparatus comprising: means for obtaining at least one image of at least one product removed from a product flow by a sorting system; means for comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and means for determining a successful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product removed from the product flow by the sorting system.

31. The apparatus of claim 30, wherein the means for obtaining at least one image of at least one product removed from a product flow by a sorting system comprises means for recording the at least one image of at least one product using a camera positioned along a reject path exiting the sorting system.

32. The apparatus of any of claims 30 to 31, wherein the means for obtaining at least one image of at least one product removed from a product flow by a sorting system comprises means for obtaining a plurality of images of a plurality of objects removed from the product flow.

33. The apparatus of any of claims 30 to 32, further comprising means for signalling the determination of the successful removal.

34. The apparatus of any of claims 30 to 33, further comprising means for obtaining at least one image of at least one product retained in the product flow by the sorting system.

35. The apparatus of claim 34, further comprising:means for comparing the at least one image of the at least one retained product with at the least one image of a foreign object in the product flow identified for removal by the sorting system; and means for determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object matches at least one image of at least one product retained in the product flow by the sorting system.

36. The apparatus of claim 35, further comprising means for signalling the determination of the unsuccessful removal.

37. The apparatus of any of claims 34 to 36, wherein the means for obtaining at least one image of at least one product retained in the product flow by a sorting system comprises means for recording the at least one image of at least one product using a camera positioned along an accept path exiting the sorting system.

38. An apparatus for confirming the removal of a foreign object from a product flow during a sorting process, the apparatus comprising: means for obtaining at least one image of at least one product removed from a product flow by a sorting system; means for comparing the at least one image of the at least one removed product with at least one image of a foreign object in the product flow identified for removal by the sorting system; and means for determining an unsuccessful removal of the foreign object from the product flow when the at least one image of the foreign object does not match the at least one image of at least one product removed from the product flow by the sorting system.

39. The apparatus of claim 38, wherein the means for determining an unsuccessful removal of the foreign object from the product flow comprises means for determining the unsuccessful removal of the foreign object from the product flow at the end of a time period.

40. The apparatus of claim 38 or claim 39, wherein the time period starts at a time of identification by the sorting system of the foreign object in the product flow.

41. A computer readable medium comprising instructions thereon which when executed by a processor cause the processor to carry out the method of any of claims 1-9 or 19-29.

42. A method substantially as herein described with reference to or as shown in any one of Figures 1 to 5.

43. An apparatus substantially as herein described with reference to or as shown in any one of Figures 1 to 5.

Images