Device and method for visual inspection

Abstract

An apparatus (10) for visual inspection, the apparatus (10) comprising a receiving component (14) configured to receive a particulate sample, and an imaging component (18) configured to capture an image of the sample. A method of visual inspection of a particulate sample is also described.

Description

Technical Field

[0001] The present invention relates to an apparatus and method for visual inspection. More particularly, the apparatus and method of the present invention are intended for visual analysis and quality inspection of grains and other particles.

[0002] More specifically, the apparatus and methods of the present invention utilize images to analyze, inspect and classify grains and other particles. Background Art

[0003] According to Grain Trade Australia (GTA) trading standards, kernels are standardised for inspection using images to identify defective grains. For the 2021 / 2022 trading standards, the Visual Identification Standards Guide 2021-2022 developed by the GTA is used for problem identification.

[0004] Currently, in order to visually evaluate grain samples, individual workers (e.g., grain samplers or assessors) must be trained to detect the presence of defective grains and manually count the defective grains. This practice provides a subjective analysis of the samples, and different samplers often provide different results, leading to inaccurate quality inspections.

[0005] In the field of grain inspection, various devices are used by samplers during the quality inspection of grain and other grains. Existing devices used in this field contain two core functions: i) mechanical dispersion of grain, and ii) capturing images of individual grain kernels for visual analysis after the grains have been mechanically dispersed.

[0006] Currently, devices used in the art to inspect grain quality utilize mechanical dispersion of grain into a channel and then feed the grain into the channel for visual analysis under a multi-spectral line scan sensor and a laser-based depth sensor. Another prior art device inspects grain quality by mechanically dispersing the grain into a line to process one grain kernel at a time and then visually analyzing the individual kernels using an arrangement of mirrors that allows multiple angled images of the individual kernels to be captured and evaluated.

[0007] Devices currently used in the art typically use mechanical dispersion to separate grains in order to capture images of individual grain kernels. This arrangement is time consuming, particularly for large samples (e.g., a half hectoliter sample of 16,000 kernels), thus reducing the overall productivity of the inspection. However, omitting mechanical dispersion would be expected to adversely affect the efficiency of standard visual inspections of grains and other particles.

[0008] The applicant has found that for quality inspection of cereals and grains it may not be necessary to capture images of individual cereal kernels and therefore mechanical dispersion in order to capture images of individual cereal kernels is not necessary.

[0009] It would be advantageous to provide a device that allows accurate, consistent and rapid visual analysis for repeatable quality inspection of grain and also reduces subjective analysis of samples, reduces instances of disparate results between different samplers and improves the accuracy of detecting the presence of defective grain.

[0010] It is an object of the apparatus of the present invention to substantially overcome the above-mentioned problems of the prior art, or at least to provide a useful alternative.

[0011] Throughout the specification, unless the context requires otherwise, the word “comprise” or variations such as “comprises” or “comprising”, will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers.

[0012] Throughout the specification, unless the context requires otherwise, the word "contain" or variations such as "contains" or "containing", will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers.

[0013] Each document, reference, patent application or patent cited in this article is expressly incorporated herein by reference, which means that the reader should read and consider it as part of this article. The contents of the documents, references, patent applications or patents cited in this article are not repeated in this article, which is only for brevity.

[0014] Reference to material or information contained in this document should not be taken as an admission that the material or information is part of the common general knowledge or is known in Australia or any other country. Summary of the invention

[0015] According to the present invention, there is provided a device for visual inspection, the device comprising:

[0016] (i) a receiving member configured to receive a particle sample; and

[0017] (ii) an imaging component configured to capture an image of the sample.

[0018] Grains include any material of similar size to cereal grains, including wheat, barley, oats and rapeseed, as well as rice, nuts and seeds. Preferably, the grain sample is a grain from an agricultural crop.

[0019] Preferably, the receiving part comprises a batch feeder configured to feed the sample onto a plate.

[0020] In one form of the invention, the batch feeder receives half a hectoliter of particles via a hopper and deposits a sample of the particles onto the tray.

[0021] The plate is preferably formed from glass.

[0022] More preferably, the receiving component further comprises one or more vibrating components configured to spread the sample on the plate. The plate is preferably vibrated in a linear manner using a vibrator.

[0023] Preferably, the imaging component comprises at least two capturing elements. The capturing elements are preferably image capturing elements.

[0024] Preferably, the image capture element is adapted to capture a view of the entirety of the sample in the dish to ensure that no defective cereals are missed.

[0025] In one form of the invention, two image capture elements are arranged in a manner such that the image capture elements are capable of capturing both a top view image and a bottom view image of the entirety of the sample in the dish. Preferably, images are captured simultaneously from each of the two image capture elements. More preferably, the background of both the top view image and the bottom view image are identical. More preferably, the background is black.

[0026] In one embodiment, the device further comprises a transfer component configured to move the sample through the device. Preferably, the transfer component comprises a conveyor. More preferably, the sample on the plate is moved through the device in a manner such that it is placed on a continuous closed loop conveyor. The conveyor preferably moves the sample through the receiving component and the imaging component.

[0027] Preferably, the device further comprises a pressurized air cleaning component.

[0028] According to another aspect of the present invention, there is provided a method for visually inspecting a particle sample, comprising the following steps:

[0029] (i) feeding the sample into a receiving part;

[0030] (ii) passing the received sample through an imaging component to capture an image of the sample; and

[0031] (iii) passing the captured image to a processing component to apply one or more data evaluation algorithms to the captured image to generate a data output.

[0032] Preferably, the data output is information relating to a physical characteristic of the sample. More preferably, the quality of the sample for visual inspection is determined from the output, including determining whether defects or contaminants are present.

[0033] In one embodiment of the method, the processing means is adapted to derive the data output using a deep learning algorithm. The deep learning algorithm allows for the separation of individual grains in the captured image of the complete sample, even if they are in contact with each other, to determine if there are any defects or contaminants in the sample. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The apparatus and method of the present invention will be described with reference to two embodiments and accompanying drawings, by way of example only, in which

[0035] Figure 1 is an exploded upper perspective view of an apparatus for visual inspection according to a first embodiment of the present invention;

[0036] Figure 2 yes Figure 1 A perspective view of a transfer component of the device;

[0037] Figure 3 yes Figure 1 An exploded perspective view of a batch feeder of the apparatus;

[0038] Figure 4 is a perspective view of a batch feeder of an apparatus according to a second embodiment of the present invention;

[0039] Figure 5 yes Figure 1 An exploded perspective view of a vibrating component of the device;

[0040] Figure 6 is an exploded perspective view of a vibrating component of an apparatus according to a second embodiment of the present invention;

[0041] Figure 7 yes Figure 1 an exploded perspective view of an imaging component of a device for capturing a top view image of a sample;

[0042] Figure 8 yes Figure 1 an exploded perspective view of an imaging component of a device for capturing a bottom-up image of a sample;

[0043] Fig. 9 is an exploded perspective view of an imaging component of an apparatus according to a second embodiment of the present invention, the imaging component being used to capture a top view image of a sample;

[0044] Fig.10is an exploded perspective view of an imaging component of an apparatus according to a second embodiment of the present invention, the imaging component being used to capture a bottom-view image of a sample;

[0045] Fig.11 is Figure 1 An image captured in an imaging component of a device, which captures a top view image of a wheat sample;

[0046] Fig.12 is Figure 1 an image captured in an imaging component of a device, which captures a bottom-up image of a wheat sample;

[0047] Fig.13 is Figure 1 an image captured in an imaging component of a device, which captures a top view image of a barley sample;

[0048] Fig.14 is Figure 1 an image captured in an imaging component of a device, which captures a bottom-up image of a barley sample;

[0049] Fig.15 is Figure 1 An image captured in an imaging component of a device capturing a top view image of an oat sample; and

[0050] Fig.16 is Figure 1 Image captured in the imaging component of the device, which captures a bottom-up image of the oat sample. DETAILED DESCRIPTION

[0051] exist Figure 1 In FIG. 1 , a device 10 for visual analysis and inspection according to a first embodiment of the present invention is shown. The device 10 comprises a receiving component 14, which is configured to receive a particle sample, and an imaging component 18, which is configured to capture an image of the sample. The particle sample is a grain, which may include wheat, barley, oats and rapeseed.

[0052] The receiving part 14 includes a batch feeder 30 configured to feed the sample onto the glass plate 12. The receiving part 14 also includes one or more vibration parts 16 configured to evenly disperse the sample on the glass plate 12. The glass plate 12 is vibrated using an exciter 50 in a linear manner.

[0053] Batch feeder 30 receives a certain amount (e.g., half a hundred liters) of grain into hopper 32 and deposits a sample thereof onto glass plate 12. Figure 3As shown, the batch feeder 30 has a hopper 32 that feeds the grain to a feed roller 34 at the bottom of the hopper 32. The feed roller 34 is configured with longitudinal grooves to receive the grain in the longitudinal grooves to ensure that when the feed roller 34 picks up the sample at the bottom of the hopper 32, there are no pinch points that may cause damage to the grain.

[0054] The rotation of the feed roller 34 may be adjusted using the motor 36 to optimize the amount of cereal dispensed onto the glass plate 12 .

[0055] According to a first embodiment of the present invention, images acquired from an imaging component (described below) are used to determine the density of the sample on the plate. The amount of grains received and rotated by the feed roller 34 can be adjusted based on the captured images to optimize the amount of grains distributed onto the glass plate 12.

[0056] exist Figure 4 , a receiving part 14 according to a second embodiment of the present invention is shown. The batch feeder 30 has a hopper 32 that feeds grain to a gate 38 at the bottom of the hopper 32. The gate 38 can be opened and closed to convey the grain into a chute 40. The volume of the chute 40 can be adjusted according to the desired volume of the grain sample. The grain in the chute 40 is then released onto the glass plate 12 by lowering a steering plate 42 at the bottom of the chute 40. The steering plate 42 is then retracted and the gate 38 is opened to receive more grain, and this process can be repeated as needed. Sample size control (or volume control) is provided by the steering plate 42 in the chute 40.

[0057] When the gate 38 is closed, grain may get stuck therein. The gate 38 includes three gate blades 44 so that when grain gets stuck, only one-third of the gate 38 can be half-open, thereby reducing the possibility of grain pouring into the chute 40. A spring mechanism 46 is attached to each gate blade 44, which provides tension to maintain the closing tendency of the gate blade 44. The spring mechanism 46 ensures that the gate blade 44 is only partially half-open, so that the grain is wedged in place. The stuck grain itself provides an obstacle, so the grain cannot flow through the partially half-open gate blade 44.

[0058] The density of the grain on the glass plate 12 affects the effectiveness of the algorithm used in the processing component (not shown), as will be described below.

[0059] According to a second embodiment of the invention, light is used to determine the density of the sample on the plate. A light source 48 above the glass plate 12 and a light sensor 63 below the glass plate 12 allow the light passing through the glass plate 12 to be measured, which indicates the density of the grains on the glass plate 12. In order to achieve the optimal density, a feedback loop may be required to adjust the amount of grains deposited.

[0060] Once the grains are deposited on the glass plate 12, they are vibrated in the vibrating element 16. Figure 5 As shown, the vibrator 50 according to the first embodiment includes a solenoid coil 52, a magnet 54, a linear shaft 56 and a linear bearing 58, which vibrate a vibrating cone 60. The solenoid coil 52 can be powered by an alternating current or waveform via an amplifier, thereby allowing the vibrator 50 to reciprocate in a linear manner, for example relative to a glass plate 12 ( Figure 5 The solenoid coil 52 is provided with a power of about 2.5 W using an alternating current of about 10 V at a frequency of 50 Hz.

[0061] In use, the vibrating cone 60 uses a spring 64 to vibrate the glass plate 12 to minimize lateral movement and ensure that the cone 60 moves in a linear manner (up and down relative to the glass plate 12 or substantially perpendicular to the glass plate 12) to evenly distribute the grains on the glass plate 12. A pair of rods 66, preferably polytetrafluoroethylene (PTFE) rods, are located between the top of the vibrating cone 60 and the glass plate 12 to transmit the vibration from the vibrating cone 60 to the glass plate 12. Two groove bearings 68 in the form of rubber wheels are located on the top side of the glass plate 12, which is to fix the glass plate 12 on the rods 66. When the vibrating component 16 is used, the glass plate 12 moves in a linear manner and the individual grain kernels collide with each other until they are randomly and generally evenly distributed. Additional vibrating components can be integrated in the receiving component 14.

[0062] like Figure 6 , a vibration exciter 50 according to a second embodiment of the present invention is shown. The vibration exciter 50 vibrates a vibration cone 60. The vibration cone 60 is attached to a frame 62 via a spring (not shown) to minimize lateral movement and ensure that the cone 60 moves in a linear manner (up and down relative to the glass plate 12), which has the additional benefit that excessive vibration is not transmitted to the frame of the conveyor 22.

[0063] The imaging component 18 includes at least two image capture elements 70 adapted to capture a view of the entirety of the sample in the glass dish 12. At least two image capture elements 70 are provided to minimize the chance that any grain defects will be overlooked, particularly for defects that may only exist on one side of the grain. By capturing images of both sides of the grain, a processing component (not shown) uses a deep learning algorithm to perform segmentation and classification, separate individual grains in the captured images, and compare at least two captured images. In this way, highly accurate visual detection of defective grains is achieved.

[0064] like Figure 1 As shown, the device 10 includes two image capturing elements 70 arranged in such a manner that two opposing image capturing elements 70 are placed above and below the glass plate 12 to capture top and bottom images of the sample.

[0065] exist Figure 7 and Figure 8 , the top and bottom image capture elements 70 are shown, respectively. The image capture element 70 includes an enclosed box 71, which also includes a camera 74, a camera mount 80, a lens 84 and a focus gear mechanism 82. The inner wall 73 of the box is covered with two rows of LED lights 72. The camera mount 80 (for example in the form of a ball and socket arrangement) is provided for precise alignment of the camera 74, and one or more servo motors are combined for focus control (including remote and automatic focus control). The focus gear mechanism 82 is integrated into the camera mount 80 and is attached to the lens 84 of each image capture element 70.

[0066] The closure box 71 is completely sealed, allowing the glass plate 75 (fixed in a slidable bracket 77 and disposed at the bottom / top of the top / bottom image capture element 70) to be easily removed for cleaning.

[0067] Fig. 9 and Fig.10 The top and bottom image capture elements 70 according to a second embodiment of the present invention are shown, respectively. The image capture element 70 is arranged as a closed box 71 having a transparent side 79 closest to the glass plate 12, the transparent side 79 being formed of non-reflective glass. The remaining inner wall 73 of the image capture element 70 is white and covered with two rows of LED lights 72. The LED lights 72 are placed to avoid light reflected from the glass plate 12 from being captured by the image capture sensor 76. To ensure a single background that facilitates image segmentation, the lens 84 of each image capture element 70 and its mount 78 on the image capture element 70 can be covered with, for example, matte fabric.

[0068] It is important that the image capture elements 70 are aligned with each other to allow the top view image and the bottom view image of the grains of cereal to accurately correspond to each other. Images are then captured simultaneously from each image capture element 70. Because the glass plate 12 is transparent, the background of the top view image is the top view of the bottom image capture element 70 itself, and vice versa, the background of the bottom view image is the bottom view of the top image capture element 70 itself.

[0069] exist Figures 11 to 16, there are shown top and bottom images of each sample selected from wheat, barley and oats, which were captured using the two top and bottom image capture elements 70 of the device 10. One of the views (either the top or bottom view) has been digitally mirrored.

[0070] The device 10 also includes a transfer component 20 configured to move the sample through the device 10. Figure 1 and Figure 2 As shown, the transfer unit 20 includes a conveyor 22, and the plurality of glass plates 12 are moved through the apparatus 10 in a continuous closed loop and stopped at each unit, such as the receiving unit 14 and the imaging unit 18, respectively. The conveyor 22 includes four rollers 24, each of which is provided with a pair of gears 26 located at its ends.

[0071] Fifteen (15) glass plates 12 are arranged on the continuous closed loop of the conveyor 22, each glass plate 12 being placed in a frame 13 in which a pair of toothed belt portions 28 are provided. Figure 2 The toothed belt portion 28 is configured to mesh with the gear 26 .

[0072] The conveyor 22 is arranged relative to the receiving part 14 and the imaging part 18 in such a manner that each individual plate 12 and frame 13 passes through the receiving part 14 and the imaging part 18 in sequence.

[0073] The device 10 may also include one or more pressurized air cleaning components, such as a high pressure blower 90, wherein an array of air nozzles directs pressurized air onto the glass plate 12. The high pressure blower 90 is positioned adjacent to the outlet tank 88 to expel any remaining debris left on the glass plate 12 into the outlet tank 88. The pressurized air cleaning component may also take the form of a blower 94 positioned adjacent to the bottom image capture element 70 to direct air for cleaning the top of the bottom image capture element 70.

[0074] The apparatus 10 may also include a ducting member (not shown) to direct the air in the apparatus 10 toward the outlet tank 88. An exhaust fan 92 that draws air out of the apparatus 10 is placed at the top of the outlet tank 88. A screen layer 93 is placed between the outlet tank 88 and the exhaust fan 92, above the lip of the outlet tank 88. The exhaust fan 92 is only activated when the apparatus is in use, and directs any light material and dust in the grain to the outlet tank 88. When the apparatus is turned off, the exhaust fan 92 is turned off, and any light material that is drawn toward the screen layer 93 will fall back into the outlet tank 88.

[0075] A method for visually inspecting a particle sample is also provided. The method includes feeding the sample into a receiving portion 12, passing the received sample through an imaging component 18 to capture an image of the sample, and transmitting the captured image to a processing component (not shown) to apply one or more data evaluation algorithms to the captured image to generate a data output.

[0076] The data output is used to determine if there are any visual defects in the grain or if there are any contaminants in the sample, such as weeds, sand, bugs or other types of grain.

[0077] The processing component (not shown) is adapted to derive a data output using a deep learning algorithm. The deep learning algorithm is used for segmentation and classification to present a data output for determining whether the grains present any visual defects or whether any contaminants are present in the sample. In detail, the processing component (not shown) separates individual grains in the captured image of the complete sample on the glass plate 12, even if they are in contact with each other.

[0078] It is conceivable that the deep learning algorithm is provided in the form of an algorithm combination combined with a large amount of training (image) data.

[0079] It is contemplated that the geometry of the hopper 32 and feed rollers 34 of the first embodiment is such that grain is retained in the hopper 32 without the need for an additional sealing mechanism.

[0080] In one embodiment of the present invention, the exciter 50 in the vibrating member 16 is a circular audio exciter. Alternative exciters may include devices that use a crank mechanism to provide linear vibrations.

[0081] In another embodiment of the present invention, the imaging component 18 may also include additional capturing elements that can capture light of different wavelengths to provide a supplement to the images of visible wavelengths (red, green, blue) captured by the image capturing element.

[0082] It is envisioned that the apparatus of the present invention will provide efficient separation of grains and achieve high throughput, thereby enabling accurate, consistent and rapid visual analysis for reproducible quality testing of the grains.

[0083] It is also contemplated that, in order to maintain accuracy and repeatability, the calibration process may be used in conjunction with a calibration plate. For example, the calibration plate would be placed as one of fifteen (15) plates, such as Figure 2 The embodiments shown in FIG. 1 and will be placed in a manner with various different objects and / or patterns to ensure that the image captured in the imaging component (top and bottom capture elements) will be within tolerances.

[0084] It should be understood that particles are not limited to cereals, but can be any material similar in size to cereal kernels, including rice, nuts and seeds.

[0085] As can be seen from the above description, the apparatus and method of the present invention provide accurate, consistent and rapid visual analysis for repeatable quality testing of grain. In addition, the apparatus and method of the present invention reduce subjective analysis of samples, thereby reducing the occurrence of different results between different samplers, and improve the accuracy of detecting the presence of defective grain.

[0086] Modifications and variations obvious to those skilled in the art are deemed to fall within the scope of the present invention.

Claims

1. A device for visual inspection, the device comprising: (i) a receiving component configured to receive a particle sample; as well as (ii) an imaging component configured to capture an image of the sample. 2 . The apparatus of claim 1 , wherein the receiving component comprises a batch feeder configured to feed the sample onto a plate.

3. The apparatus of claim 2, wherein the batch feeder receives half a hectoliter of particles via a hopper and deposits a sample of the particles onto the plate.

4. The device according to any one of claims 2 and 3, wherein the plate is formed of glass.

5. The device according to any one of the preceding claims, wherein the receiving component further comprises one or more vibrating components.

6. The apparatus of claim 5, wherein the vibrating component is configured to spread the sample on the plate.

7. The apparatus according to any one of claims 5 and 6, wherein the plate is vibrated in a linear manner using an exciter.

8. The device according to any of the preceding claims, wherein the imaging component comprises at least two capturing elements.

9. The device of claim 8, wherein the capture element is an image capture element.

10. The device according to claim 9, wherein two image capture elements are arranged in a manner such that the image capture elements can capture top view images and bottom view images of the entirety of the sample in the dish.

11. The apparatus of claim 10, wherein the images are captured simultaneously from each of the two image capture elements.

12. The device according to any one of claims 10 and 11, wherein the backgrounds of the top view image and the bottom view image are the same.

13. The device of claim 12, wherein the background is black.

14. The device of any preceding claim, further comprising a transfer component configured to move the sample through the device.

15. The device according to claim 14, characterized in that The transfer component includes a conveyor.

16. The apparatus according to any one of claims 14 and 15, wherein the samples on the plates are moved through the apparatus in a manner such that they are placed on a continuous closed loop conveyor.

17. The apparatus of any one of claims 15 and 16, wherein the conveyor moves the sample through the receiving component and the imaging component.

18. Apparatus according to any preceding claim, further comprising one or more pressurised air cleaning components.

19. A method for visually inspecting a particle sample, comprising the steps of: (i) feeding the sample into a receiving part; (ii) passing the received sample through an imaging component to capture an image of the sample; as well as (iii) passing the captured image to a processing component to apply one or more data evaluation algorithms to the captured image to generate a data output.

20. The method of claim 19, wherein the data output is information related to a physical characteristic of the sample.

21. The method of any one of claims 19 and 20, wherein the quality of the sample for visual inspection is determined from the output, including determining whether defects or contaminants are present.

22. A method according to any one of claims 19 to 21, wherein the processing component is adapted to derive the data output using a deep learning algorithm.

23. The method of claim 22, wherein the deep learning algorithm allows for separation of individual grains in the captured images of the sample.

24. A method for visually inspecting a particle sample, the method comprising the steps of: (i) feeding the sample into a device according to any one of claims 1 to 18 to capture an image of the sample; and (ii) passing the captured image to a processing component to apply one or more data evaluation algorithms to the captured image to generate a data output.

Images