Inspection device and sorting device

Abstract

To improve performance of an inspection device or a sorting device by increasing information obtained by an optical sensor.SOLUTION: An inspection device for inspecting a condition of an object includes a transfer part structured so as to transfer the object, an area sensor structured so as to image the object being transferred by action of the transfer part at a frame rate by which the same object can be imaged a plurality of times, and an identification part structured so as to identify the condition of the object on the basis of an image obtained by the area sensor.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to optical inspection and sorting techniques. [Background technology]

[0002] Optical inspection devices (hereinafter, also referred to simply as inspection devices) that identify the condition of an object (e.g., whether or not it is defective) using optical information obtained by an optical sensor when a light source is irradiated onto the object being transported have been conventionally known. For example, the following Patent Document 1 discloses a sorting device equipped with such an inspection device. This sorting device is configured to capture images of objects that have slipped off a chute with a line sensor, and to remove objects identified as defective by spraying air. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2000-197855 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, since the information obtained by the line sensor is relatively small, the conventional inspection or sorting device has room for improvement. Therefore, it is expected to increase the information obtained by the optical sensor, thereby improving the performance of the inspection or sorting device. [Means for solving the problem]

[0005] The present invention has been made to solve at least part of the above-mentioned problems, and can be realized, for example, in the following forms.

[0006] According to a first aspect of the present invention, there is provided an inspection device for inspecting a state of an object, the inspection device including a transport unit configured to transport the object, an area sensor configured to capture images of the object being transported by the action of the transport unit at a frame rate capable of capturing an image of the same object multiple times, and an identification unit configured to identify the state of the object based on the images captured by the area sensor.

[0007] The "object being transported by the action of the transport unit" includes, for example, an object being transported on the transport unit, and an object that has been transported by the transport unit and is now in the air away from the transport unit (for example, an object falling from the transport unit, an object released horizontally or diagonally upward by the transport unit). The electromagnetic wave irradiation source may irradiate at least one of visible light, near-infrared light, and X-rays.

[0008] According to this inspection device, multiple images of the same object can be obtained by the area sensor. Therefore, compared to conventional inspection devices that use a line sensor, the amount of information that can contribute to improving the identification performance can be increased. For example, when multiple images in which different areas of the same object are each captured can be obtained by the area sensor, the identification accuracy can be improved. The first embodiment can be combined with any of the following embodiments as long as no inconsistency occurs.

[0009] According to a second aspect of the present invention, there is provided an inspection device for inspecting a state of an object. The inspection device includes a transport unit configured to transport the object, an area sensor configured to capture an image of the object being transported by the action of the transport unit at a frame rate at which the same object can be captured multiple times, an identification unit configured to identify the state of the object based on the image acquired by the area sensor, and a calculation unit configured to calculate behavior information of the same object between multiple captures based on multiple images obtained by capturing the same object multiple times. The behavior information includes at least one of size change information related to a change in size of the same object between multiple images, and vector information representing at least one of a position change and a posture change of the same object.

[0010] According to this inspection device, behavior information can be obtained based on a plurality of images obtained by capturing the same object multiple times. Such information can be utilized to improve the performance of the inspection device. For example, based on such information, it can be understood whether the object is being transferred as intended (in other words, whether the inspection device is operating normally). If the inspection device is not operating as intended, the performance of the inspection device can be improved by taking necessary measures such as adjusting the equipment. Alternatively, since the shape and size of the object affect the behavior of the object, the degree of variation in the shape and size of the object can be understood based on the behavior information. The vector information may represent at least one of a position change and an attitude change in a direction different from the intended transfer direction of the object.

[0011] According to a third aspect of the present invention, there is provided a sorting device. The sorting device includes the inspection device according to the second aspect, a sorting unit configured to selectively inject air toward a specific object in accordance with the timing at which the specific object, determined based on the identification result of the identification unit, passes through a sorting area, thereby changing the transfer trajectory of the specific object, thereby sorting the specific object, and an injection control unit configured to determine an air injection mode based on at least one of size change information and vector information. According to this aspect, appropriate air injection is performed according to the behavior of the object during transfer, thereby improving sorting performance.

[0012] According to a fourth aspect of the present invention, in the third aspect, the area sensor has an angle of view capable of capturing both an object located upstream of the sorting area and an object located downstream of the sorting area on the transport path of the object, and is configured to acquire an upstream image in which an object located upstream of the sorting area is captured, and a downstream image in which an object located downstream of the sorting area is captured. The identification unit is configured to identify the state of the object based on the upstream image. Note that each of the upstream image and the downstream image may be a partial area of ​​an image in which both the upstream side and the downstream side are captured. According to this aspect, the area sensor for acquiring the upstream image to be used for the processing of the identification unit is utilized (in other words, without introducing an additional area sensor), and the downstream image is also additionally acquired, so that the situation downstream of the sorting area can be grasped. The situation downstream of the sorting area can be utilized to improve the performance of the sorting device.

[0013] According to the fifth aspect of the present invention, in the fourth aspect, the sorting device includes a monitoring unit configured to monitor the operation status of the air injection by the sorting unit based on the downstream side image. According to this aspect, it is possible to monitor whether the sorting unit is operating normally. Therefore, it is possible to prevent the sorting device from continuing to operate despite the sorting unit malfunctioning, and to prevent a deterioration in sorting performance.

[0014] According to a sixth aspect of the present invention, there is provided a sorting device for sorting objects. The sorting device includes a transport unit configured to transport objects, an area sensor configured to capture images of the objects being transported by the action of the transport unit at a frame rate capable of capturing images of the same object multiple times, an identification unit configured to identify the state of the object based on the image captured by the area sensor, and a sorting unit configured to selectively inject air toward the specific object in accordance with the timing at which the specific object determined based on the identification result of the identification unit passes through the sorting area, thereby changing the transport trajectory of the specific object, thereby sorting the specific object. The area sensor has an angle of view capable of capturing images of both the object located upstream of the sorting area and the object located downstream of the sorting area on the transport path of the object. According to this sorting device, the same effect as that of the fourth aspect can be obtained.

[0015] According to a seventh aspect of the present invention, in any one of the third to fifth aspects, the sorting device includes a prediction unit that predicts, based on at least one of size change information and vector information, at least one of the posture of the specific object when the specific object passes through the sorting area and the position of the specific object in a cross direction that crosses the intended transport direction when the specific object passes through the sorting area. The ejection control unit is configured to determine at least one of the air ejection mode and the ejection position based on the prediction result of the prediction unit. According to this aspect, sorting by ejecting air can be performed with high accuracy based on the prediction result of at least one of the posture of the specific object when the specific object passes through the sorting area and the position of the specific object in the cross direction.

[0016] According to an eighth aspect of the present invention, in the seventh aspect, when rotation of the specific object is detected based on at least one of the size change information and the vector information, the prediction unit is configured to predict the rotation angle position of the specific object when the specific object passes through the sorting area. The ejection control unit is configured to determine at least one of the air ejection mode and the ejection position based on the predicted rotation angle position. According to this aspect, when the object has a shape having a longitudinal direction and a lateral direction, at least one of the air ejection mode and the ejection position can be appropriately determined according to the rotation angle position of the specific object when the specific object passes through the sorting area (in other words, in which direction the specific object passes through the sorting area in a relatively long posture).

[0017] According to a ninth aspect of the present invention, in the seventh or eighth aspect, the behavior information includes vector information. When a position change of the specific object in the cross direction is detected based on the vector information, the prediction unit is configured to predict the position of the specific object in the cross direction when the specific object passes through the sorting area. The ejection control unit is configured to determine at least one of an air ejection mode and an ejection position based on the predicted position of the specific object in the cross direction. According to this aspect, even if the specific object is transferred in a direction other than the intended transfer direction, the position of the specific object in the cross direction when the specific object passes through the sorting area can be predicted. Therefore, based on the prediction result, at least one of an air ejection mode and an ejection position for accurately sorting the specific object can be appropriately determined.

[0018] According to a tenth aspect of the present invention, in any one of the seventh to ninth aspects, the behavior information includes vector information. The vector information includes an intended transport direction of the object. The prediction unit is configured to predict at least one of the posture of the specific object when the specific object passes through the sorting area and the position of the specific object in the cross direction when the specific object passes through the sorting area, based on the transport speed of the specific object in the intended transport direction calculated based on the vector information. According to this aspect, the time it takes for the object to move from the imaging position to the sorting area can be accurately calculated based on the actual measurement value of the transport speed of the object in the intended transport direction. Therefore, the prediction accuracy of at least one of the posture and the position in the cross direction of the specific object can be improved, and at least one of the air injection mode and the injection position can be more appropriately determined.

[0019] According to an eleventh aspect of the present invention, in the third aspect, the sorting device includes an additional area sensor configured to capture an image of an object located downstream of the sorting region on the transport path of the object. According to this aspect, the situation downstream of the sorting region can be grasped. The situation downstream of the sorting region can be utilized to improve the performance of the sorting device. The eleventh aspect may be combined with at least a part of the seventh to tenth aspects.

[0020] According to a twelfth aspect of the present invention, in the fifth aspect or in any of the seventh to tenth aspects including the fifth aspect, the downstream image includes a plurality of downstream images captured a plurality of times of the same object. The calculation unit is configured to calculate vector information of the same object based on the plurality of downstream images. The monitoring unit monitors the operation status of the air injection based on the vector information based on the plurality of downstream images. Since the transport speed and transport direction (and / or posture) of a specific object change drastically before and after air is injected, according to this aspect, the operation status of the air (i.e., whether or not the injection is performed normally) can be accurately monitored.

[0021] According to the thirteenth aspect of the present invention, in any one of the fourth, fifth, sixth, eleventh and twelfth aspects, the identification unit is configured to identify at least one of the states of an object whose transfer trajectory has been changed by the injection of air and an object whose transfer trajectory has not been changed without the injection of air, based on the downstream image. According to this aspect, in contrast to a conventional sorting device that can only grasp the state of the object before the sorting process, the state of the object discharged from the sorting device can be accurately grasped. In other words, the state of the object whose transfer trajectory has been changed can be identified in distinction from other objects. Alternatively, the state of the object whose transfer trajectory has been changed can be identified in distinction from other objects.

[0022] According to a 14th aspect of the present invention, in any one of the 4th to 6th and 11th to 13th aspects, the sorting device includes an identification processing unit configured to determine the identity of an object on the upstream side image and an object on the downstream side image. According to this aspect, it is possible to grasp whether a specific object has been accurately sorted as intended.

[0023] According to a 15th aspect of the present invention, in the 14th aspect, the identification unit is configured to identify at least one of a state of an object whose transfer trajectory has been changed by the injection of air and a state of an object whose transfer trajectory has not been changed without the injection of air, based on the upstream side image, the downstream side image, and the determination result by the identification processing unit. According to this aspect, the same effect as the 13th aspect can be obtained.

[0024] According to a sixteenth aspect of the present invention, there is provided a sorting device for sorting objects, the sorting device including: a transport unit configured to transport objects, an image sensor configured to capture an image of the objects being transported by the action of the transport unit, a recognition unit configured to identify a state of the objects based on an image captured by the image, a sorting unit configured to selectively inject air toward the specific objects in accordance with a timing at which the specific objects determined based on a result of recognition by the recognition unit pass through a sorting area, thereby changing a transport trajectory of the specific objects, thereby sorting the specific objects, and an area sensor configured to capture an image of objects located downstream of the sorting area on a transport path of the objects.

[0025] This sorting device makes it possible to grasp the situation downstream of the sorting area (e.g., the operating status of the air injection, whether a particular object is being sorted accurately as intended, the condition of the objects discharged from the sorting device, etc.). The situation downstream of the sorting area can be utilized to improve the performance of the sorting device. The sixteenth embodiment can be combined with any of the above embodiments as long as no inconsistencies arise. [Brief description of the drawings]

[0026] [Figure 1] FIG. 1 is a schematic diagram showing a schematic configuration of a sorting device according to a first embodiment. [Diagram 2] FIG. 4 is a schematic diagram showing the positional relationship between a chute and a nozzle. [Diagram 3] FIG. 2 shows an example of the same object in two images, which is the basis for calculating size change information; [Figure 4] FIG. 11 is a diagram illustrating an example of vector information. [Diagram 5] FIG. 11 is a diagram illustrating an example of vector information. [Figure 6] 10A to 10C are schematic diagrams showing an example of a method for determining an air injection mode and an injection position based on a predicted result of the behavior of an object. [Figure 7] FIG. 5 is a schematic diagram showing a schematic configuration of a sorting device according to a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] FIG. 1 is a schematic diagram showing a schematic configuration of an optical sorting device (hereinafter simply referred to as a sorting device) 10 according to a first embodiment. In this embodiment, the sorting device 10 is used to identify the state (in other words, the quality) of rice grains (more specifically, brown rice or polished rice) as an example of a sorting object (hereinafter simply referred to as an object) 90, and to sort the object 90 into good and defective products based on the result. Defective products may include, for example, broken rice, immature grains, colored grains, damaged grains, dead rice, foreign objects (for example, pebbles, mud, glass fragments, etc.). However, the object 90 is not limited to brown rice or polished rice, and may be any granular object. For example, the object 90 may be unhulled rice, wheat grains, beans (soybeans, chickpeas, edamame, etc.), resin (pellets, etc.), rubber fragments, etc.

[0028] As shown in FIG. 1, the sorting device 10 includes a detection unit 20, a storage tank 71, a feeder 72, a chute 73, a non-defective product discharge gutter 74, a defective product discharge gutter 75, a sorting section 60, and a controller 80. The controller 80 controls the overall operation of the sorting device 10. The controller 80 also functions as a recognition section 81, a calculation section 82, a jet control section 83, a monitoring section 84, a prediction section 85, and an identification processing section 86. The functions of the controller 80 may be realized by a CPU executing a predetermined program, or may be realized by a dedicated circuit (e.g., PLD, ASIC, etc.), or may be realized by a combination of a CPU and a dedicated circuit. In addition, the functions of the controller 80 may be assigned to one integrated device, or may be assigned to multiple devices in a distributed manner. Details of these functions of the controller 80 will be described later.

[0029] The storage tank 71 temporarily stores the object 90. The feeder 72 supplies the object 90 stored in the storage tank 71 onto a chute 73, which is an example of a transfer section for transferring the object. The object 90 supplied onto the chute 73 slides downward on the chute 73 and falls from the lower end of the chute 73. The chute 73 has a predetermined width (width in the chute width direction D2 shown in FIG. 2) that allows a large number of objects 90 to fall simultaneously. The object 90 released into the air from the lower end of the chute 73 falls generally in the inclined direction of the chute 73 (is transferred through the air). The direction in which the object 90 falls at this time is also referred to as the transfer direction D1 of the object 90.

[0030] In this embodiment, the object 90 is intended to be transported straight (in a direction perpendicular to the chute width direction D2) without rotating. A rotational force may be applied to the object 90 when the object 90 is released from the feeder 72, but the rotation of the object 90 caused by such a rotational force is eliminated or reduced by sliding on the chute 73. However, while the object 90 is transported in the air, due to collisions between the objects 90, the position of the object 90 may change in the chute width direction D2, which is a direction opposite to the intention, in addition to the intended transport direction D1, and the object 90 may rotate due to collisions between the objects 90. In addition, the object 90 may be transported so as to have a velocity component perpendicular to the sliding surface of the chute 73 by colliding with the bottom surface of the chute 73 and bouncing.

[0031] In an alternative embodiment, a conveyor may be used as the transport part instead of the chute 73. Alternatively, the chute 73 may be omitted. In this case, the feeder 72 serves to release the objects 90 into the air and functions as the transport part.

[0032] The detection unit 20 irradiates light onto the object 90 that has slid down the chute 73 (i.e., the object 90 falling from the chute 73) and detects light associated with the object 90 (specifically, transmitted light that has passed through the object 90 and / or reflected light that has been reflected by the object 90). In an alternative embodiment, light may be irradiated onto the object 90 sliding down the chute 73. In addition, when a conveyor is used instead of the chute 73, light may be irradiated onto the object 90 being transported on the conveyor or onto the object 90 falling from the conveyor. In a further alternative embodiment, light may be irradiated onto the object 90 that has been released horizontally or diagonally upward by a transport unit (e.g., a chute having a lower end angled horizontally or diagonally upward, or a supply device that blows the object in a desired direction with air, etc.).

[0033] 1, the detection unit 20 includes a first light source 30 and a second light source 31 as an example of an electromagnetic wave irradiation source, and an area sensor 40. The first light source 30 and the area sensor 40 are disposed on one side (also called the front side) of the transfer path (in other words, the falling trajectory) of the object 90. On the other hand, the second light source 31 is disposed on the other side (also called the rear side) of the transfer path of the object 90.

[0034] In this embodiment, each of the light sources 30 and 31 is a light source unit for irradiating the target 90 being moved (i.e., falling) with visible light. The first light source 30 emits a first light 32. Similarly, the second light source 31 emits a second light 33. Each of the lights 32 and 33 has a wavelength corresponding to red, a wavelength corresponding to green, and a wavelength corresponding to blue. In this embodiment, each of the light sources 30 and 31 includes a so-called color LED. Each of the light sources 30 and 31 is a line light source in which a plurality of LEDs are arranged in the chute width direction D2. However, the specifications of the light sources 30 and 31 (e.g., the number, the light emission type, the wavelength region, etc.) are not particularly limited. For example, each of the light sources 30 and 31 may be an area light source in which a plurality of LEDs are arranged two-dimensionally.

[0035] The area sensor 40 is an optical sensor capable of detecting red light, green light, and blue light individually. The area sensor 40 has a plurality of photodetection elements arranged two-dimensionally (in this embodiment, along the conveying direction D1 and the chute width direction D2). In this embodiment, the area sensor 40 is a CCD sensor, but may be another type of sensor such as a CMOS sensor.

[0036] The area sensor 40 detects light associated with the object 90 being moved. Specifically, the area sensor 40 arranged on the front side can detect a first light 32 (hereinafter also referred to as reflected light 32) emitted from a first light source 30 on the front side and reflected by the object 90, and a second light 33 (hereinafter also referred to as transmitted light 33) emitted from a second light source 31 on the rear side and transmitted through the object 90.

[0037] The type of light detected by the area sensor 40 is determined by the lighting pattern of the light sources 30 and 31. In a first lighting pattern in which the first light source 30 and the second light source 31 are simultaneously turned on, the area sensor 40 detects light (hereinafter also referred to as reflected and transmitted light) that is a combination of the reflected light 32 and the transmitted light 33. In a second lighting pattern in which the first light source 30 is turned on and the second light source 31 is turned off, the area sensor 40 detects only the reflected light 32. In a third lighting pattern in which the first light source 30 is turned off and the second light source 31 is turned on, the area sensor 40 detects only the transmitted light 33. Hereinafter, at least one of the reflected light, transmitted light, and reflected and transmitted light described above is also referred to as light associated with the object 90.

[0038] Which of the first to third lighting patterns is to be adopted can be arbitrarily determined depending on the type and properties of the object 90 and the type of defective product to be removed. Only one of the first to third lighting patterns may be adopted. Alternatively, two or more of the first to third lighting patterns may appear alternately at a predetermined time interval or according to a predetermined repetition rule.

[0039] The specifications of the area sensor 40 (e.g., the number, the installation position, the spectral sensitivity characteristics, etc.) are not particularly limited and can be arbitrarily determined according to the specifications of the light sources 30 and 31. For example, a single area sensor 40 may be disposed only on the rear side. Alternatively, two area sensors 40 may be disposed on the front side and the rear side, respectively.

[0040] The output from the area sensor 40, i.e., an analog signal representing the intensity of the detected light, is converted to a digital signal by an AC / DC converter (not shown). This digital signal is input to the controller 80 as image data. The controller 80 determines the state of the object 90 based on the input image. Such a determination is performed for each of the objects 90 as processing by the recognition unit 81.

[0041] In this embodiment, the state determined by the identification unit 81 includes a color state (in other words, an optical state) and a shape and / or a size state. The state also includes a feature amount represented by a physical quantity and a quality determined based on the feature amount.

[0042] The characteristic amount of the color state includes a color gradation value of each pixel of the image representing the object 90. The characteristic amount of the shape and / or size state may include, for example, at least one of the area, height, width, perimeter length, and circularity of the whole and / or part of the object 90.

[0043] In this embodiment, "quality" includes, for example, a classification between good products (i.e., rice grains of relatively high quality) and defective products (i.e., rice grains of relatively low quality and / or foreign objects). However, "quality" may also include the type of defect (e.g., whether it corresponds to broken rice, immature grains, colored grains, damaged grains, dead rice, or foreign objects). Alternatively, "quality" may include a classification between objects that should be removed in the sorting unit 60 and objects that should not be removed. Furthermore, "quality" includes quality determined based on the color state and quality determined based on the shape and / or dimensional state. Defective products determined based on the color state may include, for example, immature grains, colored grains, damaged grains, dead rice, and foreign objects. Defective products determined based on the shape and / or dimensional state may include, for example, broken rice, insect-damaged grains, and foreign objects.

[0044] In this embodiment, the identification unit 81 determines whether the object 90 is a good product or a defective product by comparing the characteristic amount of the color state (in other words, the gradation value of the image data) with a predetermined threshold value (in other words, based on whether the characteristic amount of the color state is within a predetermined normal range). Such a determination may be made based on a representative value (e.g., an average value, a median value, a maximum value, a minimum value, etc.) of the gradation values ​​of a plurality of pixels constituting the image of the object 90. Alternatively, the defective product may include an object 90 having a partial defect of a predetermined size or more. Such a partial defect may be determined based on the criterion that the number of pixels whose gradation values ​​are not within the normal range among the plurality of pixels constituting the image of the object 90 is a predetermined number or more (in other words, the area of ​​the defective part is a predetermined value or more).

[0045] Furthermore, in this embodiment, the identification unit 81 determines whether the object 90 is a good or bad product by comparing the features of the shape and / or dimensional condition with a predetermined threshold value (in other words, based on whether the shape and / or dimensional features are within a predetermined normal range).

[0046] In this embodiment, the frame rate (unit: Hz or frame / sec) and the size of the field of view (height in the transport direction D1) of the area sensor 40 are set so that the same object 90 can be imaged multiple times. The identification unit 81 then identifies the state of the object 90 based on multiple images obtained by imaging the same object 90 multiple times. For this reason, when the object 90 is transported in the air while changing its posture, the area of ​​the same object 90 that is imaged at least once is expanded. As a result, the identification accuracy by the identification unit 81 is improved. However, the identification unit 81 may identify the state of the object 90 based on a single image.

[0047] Identification of the same object 90 between the multiple images may be performed by any known method as a process of the identification processing unit 86. For example, known methods such as template matching and optical flow estimation methods may be used. Alternatively, when the frame rate is relatively high, overlapping between the multiple objects 90 in the first image and the multiple objects 90 in the second image may be detected, and the overlapping objects 90 may be determined to be the same object 90. Alternatively, the center of gravity of each of the multiple objects 90 in the first image and the center of gravity of each of the multiple objects 90 in the second image may be calculated. In this case, when the center of gravity of each of the multiple objects 90 in the first image is shifted by a moving distance (a distance assumed in advance) corresponding to the frame rate, the centers of gravity that are the smallest apart between the first image and the second image may be determined to be the centers of gravity of the same object 90.

[0048] The sorting unit 60 sorts the objects 90 based on the state determined by the identification unit 81. This sorting is performed by a trajectory changing operation for changing the trajectory of a specific object 90. Specifically, as shown in Fig. 1, the sorting unit 60 includes a plurality of nozzles 61 and a plurality of valves 63. In Fig. 1, for the sake of simplicity, the number of nozzles 61 and the number of valves 63 are shown to be the same, but the correspondence between the number of nozzles 61 and the number of valves 63 actually depends on the number of openings that the nozzles 61 have.

[0049] FIG. 2 is a schematic diagram showing the positional relationship between the chute 73 and the nozzles 61. As shown in the figure, the multiple nozzles 61 are arranged in the chute width direction D2. In this embodiment, each of the nozzles 61 has multiple openings 62 arranged in the chute width direction D2. The above-mentioned valve 63 (not shown in FIG. 2) is arranged for each of the multiple openings 62. The position of each of the openings 62 in the chute width direction D2 is previously associated with the position of the image acquired by the area sensor 40 in the direction corresponding to the chute width direction D2. Each of the openings 62 of the nozzles 61 is connected to a compressor (not shown) via the corresponding valve 63.

[0050] Returning to FIG. 1 for the explanation, the sorting unit 60 performs a trajectory change operation by selectively injecting air 64 from the opening 62 of the nozzle 61 toward the specific object 90 in accordance with the timing of passing through the sorting area 65. Specifically, one or more valves 63 are selectively opened in response to a control signal from the controller 80, and air 64 is injected toward the specific object 90 from one or more openings 62 corresponding to the opened valve 63. The sorting area 65 is an area on the transfer path (falling trajectory) of the object 90 through which the air 64 injected from the nozzle 61 passes. In this embodiment, the specific object 90 to which the air 64 should be injected is an object 90 having a predetermined defect (more specifically, the specific object 90 is an object 90 having at least one of a color defect and a shape and size defect). However, the specific object 90 to which the air 64 should be injected can be set arbitrarily. For example, the air 64 may be injected only toward the object 90 having a color defect and a shape and size defect. Alternatively, air 64 may be sprayed against the object 90 that has been determined to be a non-defective product (so-called reverse spraying).

[0051] The timing at which the object 90 passes through the sorting area 65 is determined by the jetting delay time. The jetting delay time is the time required from when the specific object 90 is imaged by the area sensor 40 until the air 64 is jetted toward the specific object 90 (in other words, the time required for the object 90 to be transported from the imaged position to the sorting area 65). The jetting delay time may be determined in advance. Alternatively, the jetting delay time may be determined based on the transport speed of the object 90 in the transport direction D1 calculated based on multiple images acquired by the area sensor 40. A method for calculating the transport speed will be described later.

[0052] Certain objects 90 are blown away by the air 64, deviate from the falling trajectory along the transfer direction D1, and are guided to the defective product discharge gutter 75 (shown as object 91 in FIG. 1). On the other hand, air 64 is not sprayed onto objects 90 that are determined to be non-defective. Therefore, objects 90 that are determined to be non-defective are guided to the non-defective product discharge gutter 74 without changing their falling trajectory (shown as object 92 in FIG. 1).

[0053] In this embodiment, the sorting device 10 is configured to grasp the behavior of the object 90 during transportation based on the image acquired by the area sensor 40, and to determine the injection mode of the air 64 based on the behavior. Specifically, the controller 80 first calculates, as the processing of the calculation unit 82, behavior information of the same object 90 between multiple imaging operations based on multiple images obtained by imaging the same object 90 multiple times. In this embodiment, this behavior information includes size change information. The size change information is information related to the change in size of the same object 90 between multiple images. Such a change in size may occur due to a change in posture and / or a bounce of the object 90.

[0054] The fact that the size of the same object 90 differs between multiple images by a predetermined degree or more may indicate peculiar behavior of the object 90. For example, if the object 90 has a flat shape like rice, the size of the same object 90 may differ significantly between multiple images due to a change in the posture of the object 90 (rotation around a rotation axis extending in a direction other than a direction perpendicular to the field of view of the area sensor 40). Alternatively, if the object 90 does not slide on the chute 73 as intended and hits the bottom surface of the chute 73 and bounces off, the size of the object 90 may differ significantly between multiple images even if there is no change in the posture of the object 90 as seen from the area sensor 40.

[0055] FIG. 3 shows an example of an image of the object 90 when such a bounce occurs. In FIG. 3, the object 90 on the image acquired by the area sensor 40 in the Nth (N is a natural number) scan is shown as an object 93, and the object 90 on the image acquired in the N+1th scan (the same object 90 as the object 93) is shown as an object 94. The pixel representing the object 90 can be easily grasped, for example, by binarizing any of the images of each color of RGB. This example shows a case where the object 90 is transported while gradually deviating toward the front side with respect to the intended transport path between the Nth scan and the N+1th scan due to the bounce of the object 90. As can be seen from FIG. 3, the object 94 is larger in size in the X direction and the Y direction than the object 93 due to the bounce of the object 90, in other words, due to the behavior of the object 90 deviating from the intended transport path in a direction intersecting the field of view of the area sensor 40. The X direction is the direction corresponding to the chute width direction D2 on the image, and the Y direction is the direction corresponding to the transport direction D1 on the image.

[0056] The size change information that can represent the behavior of the object 90 as described above is typically the area ratio of the same object 90 between multiple images. However, the size change information can be any index that can represent the degree of change in size of the object 90. For example, instead of or in addition to the area ratio, the ratio of the perimeter lengths may be adopted. Alternatively, instead of or in addition to the ratio, the difference may be adopted.

[0057] After calculating the size change information, the controller 80, as processing by the ejection control unit 83, determines the ejection mode of the air 64 by the sorting unit 60 (in other words, how to eject the air 64 against a specific target 90) based on the size change information. This ejection mode can be any parameter whose setting varies the range of influence of the ejected air 64 on the target 90 (the range in which the force of the air 64 can act on the target 90 to such an extent that the trajectory of the target 90 can be changed). For example, the ejection mode may be one or a combination of two or more of the ejection range of the air 64 in the chute width direction D2, the ejection period (in other words, the time that the ejection continues), and the ejection pressure.

[0058] The spray range of the air 64 in the chute width direction D2 is determined by how many consecutive openings 62 among the adjacent openings 62 from which the air 64 is sprayed simultaneously. The number of openings 62 from which the air 64 is sprayed simultaneously may be one or more than one. For example, the technique described in JP 2022-115568 A may be used to control the spray range of the air 64 in the chute width direction D2. According to this technique, the size of the spray range is controlled by variably setting the presence or absence of overlap of the spray range on the image assigned to each of the multiple openings 62, or the degree of overlap. JP 2022-115568 A is incorporated herein by reference in its entirety.

[0059] The ejection period affects the ejection range of the object 90 in a coordinate system that moves together with the object 90 being moved. Specifically, the longer the ejection period is set, the larger the range of influence of the air 64 on the object 90 in the moving direction D1 becomes. The ejection period is controlled by controlling the time for which the valve 63 is kept open.

[0060] The spray pressure is one of the parameters that correlates with the range of influence of the air 64. Specifically, the higher the spray pressure of the air 64, the wider the range of influence of the air 64 becomes, since the sprayed air 64 spreads over a wider area on the plane defined by the transfer direction D1 and the chute width direction D2. The spray pressure may be controlled, for example, by controlling the opening of the valve 63.

[0061] In this embodiment, if the degree of size change represented by the size change information is equal to or greater than the first threshold, the controller 80, as processing by the spray control unit 83, determines the spray mode of the air 64 so that the range of influence of the air 64 is larger than when the degree of size change is less than the first threshold. In other words, if the degree of size change represented by the size change information is equal to or greater than the first threshold, the controller 80 determines at least one of the spray range, spray period, and spray pressure so that at least one of the following is satisfied: the spray range of the air 64 in the chute width direction D2 is wider, the spray period is longer, and the spray pressure is higher.

[0062] According to this configuration, the sorting performance can be improved by performing appropriate air injection according to the behavior of the object 90 during transfer. Specifically, the fact that the degree of change in size indicated by the size change information is equal to or greater than the first threshold value means that the object 90 is rotating and / or bouncing, as described above. Such behavior of the object 90 can cause a large deviation between the expected (designed) value and the actual value (i.e., the intended sorting fails) with respect to at least one of the timing when the specific object 90 reaches the sorting area 65 and the position in the chute width direction D2 when the specific object 90 reaches the sorting area 65. However, according to the above configuration, when such behavior of the object 90 is detected, control is performed to increase the range of influence of the air 64, so that the specific object 90 can be reliably sorted. In addition, since it is not necessary to always set the injection mode of the air 64 on the safe side so that the specific object 90 can be reliably sorted no matter what behavior the specific object 90 exhibits, it is possible to achieve both improved yield and improved sorting accuracy.

[0063] In an alternative embodiment, the behavior information calculated by the calculation unit 82 includes vector information instead of or in addition to the size change information. The vector information is information that represents at least one of a position change and a posture change in a predetermined direction of the same object 90, and can be expressed as a vector. In this embodiment, the predetermined direction includes the intended transfer direction of the object 90 (i.e., the transfer direction D1) and a direction different from the transfer direction D1. In addition, such a direction different from the transfer direction D1 includes the chute width direction D2 and the rotation direction. This rotation direction is defined as the direction of rotation on the field of view of the area sensor 40 when the object 90 is transferred while rotating. However, the predetermined direction can be set arbitrarily, and for example, the predetermined direction may include only a direction different from the transfer direction D1. In addition, the direction different from the transfer direction D1 may be only one of the chute width direction D2 and the rotation direction.

[0064] 4 and 5 show examples of vector information calculated by the calculation unit 82. In FIG. 4 and FIG. 5, similarly to FIG. 3, the object 90 on the image acquired in the Nth scan is shown as the object 93, and the object 90 (the same object 90 as the object 93) on the image acquired in the N+1th scan is shown as the object 94. In the example of FIG. 4, the direction and distance that the same object 90 has moved during the time required for one scan are shown as vector information 95. In this example, the vector information 95 is obtained by linearly connecting the center of gravity 93a of the object 93 and the center of gravity 94a of the object 94. The coordinate values ​​of the centers of gravity 93a and 94a may be expressed with a pixel-by-pixel resolution or may be expressed with a higher resolution. The coordinate value of the center of gravity 93a can be calculated, for example, by dividing the sum of the coordinate values ​​of each coordinate point corresponding to the target object 93 for each of the X coordinate (this is the coordinate in the X direction corresponding to the chute width direction D2) and the Y coordinate (this is the coordinate in the Y direction corresponding to the transfer direction D1) by the number of coordinate points. The coordinate value of the center of gravity 94a can be calculated in a similar manner.

[0065] 4, it can be seen that the same object 90 moves a distance L1 in the transfer direction D1 between the two images, and also changes in position (lateral shift) a distance L2 in the chute width direction D2. The transfer speeds of the object 90 in the transfer direction D1 and the chute width direction D2 can be calculated by dividing the distances L1 and L2 by the time required for one scan of the area sensor 40. The transfer speed in the transfer direction D1 calculated in this way may be used to determine the injection delay time, as described above.

[0066] 5, the direction and distance that the same object 90 has moved during the time required for one scan are shown as vector information 96. The vector information 96 is obtained by connecting the center of gravity 93a of the object 93 and the center of gravity 94a of the object 94 in a straight line. In this example, no lateral displacement has occurred in the object 90.

[0067] However, in the example of FIG. 5, the posture of the object 90 changes during the time required for one scan. This posture change is represented by vector information 97. The vector information 97 can also be said to be vector information related to the rotation of the object 90. In this embodiment, the vector information 97 is calculated based on the angle change of the long axis of the same object 90. Specifically, first, the long axes 93b and 94b of the objects 93 and 94 are detected. The long axis 93b is obtained, for example, by linearly connecting two pixels (or two coordinate values) with the greatest separation distance among combinations of two pixels (or two coordinate values) arbitrarily selected from each pixel (or coordinate value with a resolution higher than the pixel unit) constituting the outer periphery (outer periphery) of the object 93. The long axis 94b is also obtained by a similar method. Then, the angle θ3 (=θ1-θ2) between the long axis 93b and the long axis 94b is calculated. This provides vector information 97 having a magnitude equal to the angle θ3 between major axis 93b and major axis 94b in the direction of rotation (the direction of rotation about the axis of rotation extending in a direction perpendicular to the field of view of area sensor 40). However, vector information 97 representing a change in posture can be calculated by any method. For example, vector information 97 may be calculated based on an angular change in the minor axis of the same object 90. Alternatively, the average value of the angular change in the major axis and the angular change in the minor axis may be treated as the magnitude of vector information 97.

[0068] In the examples of Fig. 4 and Fig. 5, the vector information is calculated based on two images obtained by two consecutive scans, but the vector information may be calculated based on two images obtained by two non-consecutive scans (e.g., the Nth scan and the N+2th scan). Alternatively, the vector information may be calculated based on three or more images. In this way, the accuracy of the vector information can be improved. In this case, the vector information may be calculated for each two consecutively acquired images that are extracted in sequence from three or more images, and the vector information obtained by averaging the obtained multiple pieces of vector information may be used.

[0069] After calculating the vector information as described above, the controller 80, as processing by the prediction unit 85, predicts, based on the vector information, the orientation of the specific object 90 when the specific object 90 passes through the sorting area 65. Furthermore, as processing by the prediction unit 85, the controller 80 predicts, based on the vector information, the position of the specific object 90 in a direction intersecting the intended transfer direction (i.e., transfer direction D1) when the specific object 90 passes through the sorting area 65 (in this embodiment, the chute width direction D2, hereinafter also referred to as the intersecting direction). In an alternative embodiment, the prediction unit 85 may predict only one of the orientation of the specific object 90 when it passes through the sorting area 65 and the position of the specific object 90 in the intersecting direction.

[0070] The posture of the specific object 90 when passing through the sorting area 65 is predicted based on the vector information 97 (i.e., vector information related to rotation) illustrated in FIG. 5. Specifically, first, the angular velocity of the specific object 90 is calculated from the vector information 97 and the scan rate of the area sensor 40. In the example of FIG. 5, a value obtained by dividing the angle θ3 by the time required for one scan is calculated as the angular velocity. Then, based on this angular velocity and the time required from the imaging position of the area sensor 40 to the sorting area 65, the posture (rotation angle position) of the specific object 90 when passing through the sorting area 65 is predicted. Since the distance from the imaging position of the area sensor 40 to the sorting area 65 is known, the time required from the imaging position of the area sensor 40 to the sorting area 65 is obtained by dividing the distance by the transfer speed of the object 90 in the transfer direction D1. The transfer speed in the transfer direction D1 may be a predetermined design value or may be an actual measurement value calculated by the above-mentioned method. Using actual measurements can improve prediction accuracy.

[0071] 4, the position of the specific object 90 in the chute width direction D2 when passing through the sorting area 65 is predicted based on vector information 95 (i.e., vector information related to a change in position). Specifically, the transfer speed of the specific object 90 in the chute width direction D2 is calculated from the X-direction component (component in the chute width direction D2) of the vector information 95 and the scan rate of the area sensor 40. Specifically, as described above, the value obtained by dividing the distance L2 by the time required for one scan is calculated as the transfer speed in the chute width direction D2. Then, the position of the specific object 90 in the chute width direction D2 when passing through the sorting area 65 is predicted based on this transfer speed in the chute width direction D2 and the time required from the imaging position of the area sensor 40 to the sorting area 65.

[0072] Next, the controller 80 determines the ejection mode of the air 64 based on the prediction result of the prediction unit 85 as a process of the ejection control unit 83. FIG. 6 is a schematic diagram showing an example of a method for determining the ejection mode of the air 64 based on the prediction result. In the figure, objects 90a to 90c, 90d to 90f, and 90g to 90i are each a transfer trajectory of the same specific object 90. Vector information 95a calculated based on the behavior of the same object 90 that moved from the position of the object 90a to the position of the object 90b has only a component in the transfer direction D1. Therefore, it is predicted that the object 90 will be transferred straight to the position of the object 90c on the sorting area 65. Therefore, the controller 80 ejects the air 64 from the opening 62b corresponding to the position of the object 90c among the openings 62a to 62i of the nozzle 61.

[0073] Vector information 95d calculated based on the behavior of the same object 90 that moves from the position of object 90d to the position of object 90e has a component in the chute width direction D2. Therefore, it is predicted that the object 90 will be moved obliquely to the position of object 90f on the sorting area 65. Therefore, the controller 80 jets air 64 from the opening 62d corresponding to the position of object 90f. In this embodiment, the jetting of air 64 from the opening 62d in this case is performed in the same manner as the jetting of air 64 to the object 90 corresponding to the vector information 95a. However, in an alternative embodiment, the jetting mode may be determined for the object 90 corresponding to the vector information 95d having a component in the chute width direction D2 so that the range of influence of the air 64 on the object 90 is increased. In other words, the jetting time may be determined to be longer, the jetting pressure may be determined to be higher, or the air 64 may be jetted from two or more consecutive openings 62 including the opening 62d.

[0074] Vector information 95g (vector information related to position change) calculated based on the behavior of the same object 90 that has moved from the position of object 90g to the position of object 90h has only a component in the transfer direction D1. Therefore, it is predicted that the object 90 will be transferred straight to the position of object 90i on the sorting area 65. Furthermore, vector information 97g (vector information related to rotation) is also detected based on the behavior of the same object 90 that has moved from the position of object 90g to the position of object 90h. In this example, based on this vector information 97, it is predicted that the object 90 will take a posture with a major axis extending in the chute width direction D2 when it is located on the sorting area 65. Therefore, the controller 80 sprays air 64 from the openings 62g and 62h corresponding to the object 90 at the position of object 90i in such a posture. That is, in this example, the number of openings 62 that spray air 64 is increased according to the posture of the object 90 predicted based on the vector information 97g. In alternative embodiments, instead of or in addition to increasing the number of openings 62 through which air 64 is injected, the injection time may be determined to be longer and / or the injection pressure may be determined to be higher.

[0075] According to the various controls described above, the posture of the specific object 90 and the position of the specific object 90 in the chute width direction D2 when passing through the sorting area 65 are predicted based on vector information. Then, based on the prediction result, the spraying mode of the air 64 and the spraying position in the chute width direction D2 are determined. Therefore, it is possible to spray air 64 appropriately according to the posture and position of the specific object 90, improving sorting accuracy.

[0076] In an alternative embodiment, the jet control unit 83 may determine the jet mode of the air 64 based on the vector information calculated by the calculation unit 82, without performing prediction by the prediction unit 85. For example, the jet control unit 83 may determine the jet mode such that the range of influence of the air 64 on the target 90 is larger when vector information relating to rotation is detected than when it is not detected. Alternatively, the jet control unit 83 may determine the jet mode such that the range of influence of the air 64 on the target 90 is larger when vector information representing a lateral deviation is detected than when it is not detected.

[0077] In a further alternative embodiment, when the orientation of the target object 90 is predicted to be vertically elongated based on the vector information regarding the rotation, the ejection control unit 83 may determine the ejection period to be relatively long. Alternatively, when the orientation of the target object 90 is predicted to be horizontally elongated, the ejection control unit 83 may determine the ejection period to be relatively short. According to this configuration, the ejection period can be appropriately set to a required extent according to the orientation of the specific target object 90, so that both sorting accuracy and yield can be achieved.

[0078] Here, the description will be returned to FIG. 1. As shown in FIG. 1, in this embodiment, the area sensor 40 has an angle of view 41 capable of capturing both the object 90 located upstream of the sorting area 65 on the transport path of the object 90 and the object 90 located downstream of the sorting area 65. In the following description, among the images acquired by the area sensor 40, an image area captured upstream of the sorting area 65 is called an upstream image, and an image area captured downstream of the sorting area 65 is called a downstream image. The above-mentioned identification process by the identification unit 81 (process of identifying the state of the object 90 for sorting by the sorting unit 60) is performed using the upstream image. The downstream image is used for a purpose different from that of the upstream image, as described below.

[0079] Specifically, the controller 80 monitors the operation status of the jetting of the air 64 by the sorting unit 60 based on the downstream side image as a process of the monitoring unit 84. In this embodiment, the controller 80 calculates vector information based on a plurality of downstream side images captured multiple times for the same object 90, and monitors the operation status of the jetting of the air 64 based on the calculated vector information. The method of calculating the vector information is as described above. Since the vector information of a specific object 90 (in other words, the transport speed and transport direction, and / or the change in posture) changes drastically before and after the air 64 is jetted, this configuration makes it possible to monitor whether the jetting of the air 64 by the sorting unit 60 is performed normally (in other words, whether any abnormality such as clogging of the air flow path of the sorting unit 60 has occurred).

[0080] The determination as to whether the air 64 has been ejected normally may be made based on, for example, whether the transport speed represented by the vector information is within a predetermined numerical range and / or whether the transport direction represented by the vector information is within a predetermined directional range. Alternatively, the determination as to whether the air 64 has been ejected normally may be made by any known method using artificial intelligence. For example, this determination may be made by applying the vector information to an inference model generated by machine learning.

[0081] In an alternative embodiment, a determination may be made based on a single downstream image without vector information as to whether air 64 was successfully sprayed, in which case the determination may be made in any manner known in the art using artificial intelligence.

[0082] When it is determined that an abnormality has occurred in the ejection operation of the air 64, the controller 80 may notify the user of the abnormality by any known method. With this configuration, it is possible to prevent the sorting device 10 from being continuously used despite the occurrence of an abnormality in the ejection operation of the air 64. Furthermore, the controller 80 may output the determination result regarding the ejection operation of the air 64 or the statistical data thereof to any device (e.g., a display, a communication interface, a storage medium, a printing device, etc.). With this configuration, this output result can be used as information for adjustment work (adjustment work for eliminating the abnormality) of the sorting unit 60. The determination result regarding the ejection operation of the air 64 may be output in association with the X-direction coordinate value of the downstream side image.

[0083] Furthermore, in this embodiment, the controller 80 performs an identification process for determining the identity of the object 90 in the upstream image and the object 90 in the downstream image as the process of the identification processing unit 86. More specifically, the controller 80 determines the identity of the object 90 determined by the identification unit 81 to be a specific object 90 based on the upstream image (i.e., the object 90 determined to be sprayed with air 64) and the object 90 determined by the monitoring unit 84 to have had the transfer trajectory changed by the spraying of air 64 based on the downstream image. The method of the identification process is as described above. Then, based on the result of the identification process (determination result of identity), the controller 80 determines whether the object 90 determined by the monitoring unit 84 to have had the trajectory changed by the spraying of air is a specific object 90 (i.e., the object 90 determined by the identification unit 81 to have had the air 64 sprayed), and outputs the result or the statistical data thereof to an arbitrary device. According to this configuration, it is possible to grasp whether the specific object 90 has been accurately selected as intended. Such information can be used, for example, to adjust the sorting unit 60 .

[0084] In addition, in this embodiment, the controller 80 associates and aggregates the identification result of the state of the object 90 based on the upstream side image and the judgment result of whether or not the trajectory has been changed based on the downstream side image (in other words, the judgment result of whether each object 90 has been guided to the defective product discharge gutter 75 or the non-defective product discharge gutter 74), and outputs the aggregation result to an arbitrary device. For example, the controller 80 outputs statistical data of the state of the object 90 using only the object 90 guided to the non-defective product discharge gutter 74 as a parameter (hereinafter referred to as non-defective product state data), and statistical data of the state of the object 90 using only the object 90 guided to the defective product discharge gutter 75 as a parameter (hereinafter referred to as defective product state data). According to this configuration, in contrast to the conventional method that relies on visual confirmation by the user, a new function of the sorting device is provided, and the user can distinguish and accurately grasp the state of the object 90 guided to the non-defective product discharge gutter 74 and the state of the object 90 guided to the defective product discharge gutter 75. In an alternative embodiment, only one of the good condition data and the bad condition data may be output.

[0085] In an alternative embodiment, the controller 80 may output at least one of the non-defective item state data and the defective item state data without determining the identity of the object 90 in the upstream side image and the object 90 in the downstream side image. For example, the controller 80 may identify the state of each object 90 based on the downstream side image as the processing of the identification unit 81. The controller 80 may then associate such identification results with the determination result of the presence or absence of a trajectory change based on the downstream side image, tally them up, and output the tally result to an arbitrary device.

[0086] According to the above-described sorting device 10, it is possible to detect peculiar behavior (unintended behavior) of the object 90. In addition, various controls can be performed to suppress a decrease in the identification accuracy and / or sorting accuracy caused by such peculiar behavior. The peculiar behavior may include, for example, the object 90 deviating from an expected transfer trajectory due to a collision between the objects 90, a collision with a component of the sorting device 10 (the chute 73 in the above-described embodiment), etc., the transfer speed of a specific object 90 being significantly different from the transfer speed of other objects 90 due to the posture, shape, weight, etc. of the object 90, the object 90 being transferred while rotating due to the shape of the object 90, a collision between the objects 90, etc.

[0087] FIG. 7 is a schematic diagram showing a schematic configuration of a sorting device 110 according to a second embodiment. Hereinafter, the sorting device 110 will be described only in terms of differences from the first embodiment. The sorting device 110 includes a first area sensor 140 and a second area sensor 145 (in other words, an additional area sensor) instead of the area sensor 40. The first area sensor 140 has a field angle 141 that can capture only the object 90 located upstream of the sorting region 65 (the object 90 located downstream of the sorting region 65 cannot be captured). The second area sensor 145 has a field angle 146 that can capture only the object 90 located downstream of the sorting region 65 (the object 90 located upstream of the sorting region 65 cannot be captured). As a result, the first area sensor 140 acquires an upstream image in which a field of view located upstream of the sorting region 65 is captured, and the second area sensor 145 acquires a downstream image in which a field of view located downstream of the sorting region 65 is captured. The sorting device 110 configured in this manner can also similarly execute the various processes described above of the sorting device 10 (various processes performed using the images acquired by the area sensor 40).

[0088] In this embodiment, the first area sensor 140 and the second area sensor 145 have different imaging directions (directions perpendicular to the field of view). This makes it possible to set the angle of view 146 so as to more easily determine the operating condition of the jet of air 64 (i.e., whether or not the trajectory of the target 90 has changed due to the jet of air 64).

[0089] In an alternative embodiment, the sorting device 110 may include a line sensor instead of the first area sensor 140 for acquiring an upstream image.

[0090] Although the embodiment of the present invention has been described above, the above embodiment is for the purpose of facilitating understanding of the present invention and does not limit the present invention. The present invention may be modified or improved without departing from the spirit thereof, and the present invention includes equivalents thereof. In addition, any combination or omission of each component described in the claims and specification is possible within the scope of solving at least a part of the above-mentioned problems or achieving at least a part of the effects.

[0091] For example, the sorting device 10, 110 may include an electromagnetic wave irradiation source that irradiates the object 90 with electromagnetic waves having any wavelength, instead of or in addition to the visible light source (light sources 30, 31 in the above example). Such electromagnetic waves may be, for example, at least one of X-rays and near-infrared rays. In this case, the sorting device 10, 110 may include, instead of or in addition to the area sensor 40, 140, 145, an area sensor having a plurality of electromagnetic wave detection elements corresponding to the wavelength of the electromagnetic waves irradiated by the electromagnetic wave irradiation source.

[0092] The present invention is not limited to a sorting device, and can be realized in various forms. For example, the present invention may be realized as an inspection device for inspecting the state of an object. Such an inspection device may have a configuration in which the sorting unit 60 and the defective product discharge gutter 75 are removed from the above-mentioned sorting device 10. In this case, the controller 80 may output behavior information (or its statistical data) calculated by the calculation unit 82 in addition to the state of the object 90 identified by the identification unit 81 to any device (for example, a display, a communication interface, a storage medium, a printing device, etc.).

[0093] Such an inspection device can confirm, based on the behavior information, whether the inspection device is operating normally so that the object 90 is transferred as intended. In addition, when the uniformity of the object's features (shape and / or size) is low, the probability of the object behaving peculiarly when transferred increases, so this inspection device can determine the uniformity of the object's features based on the behavior information. [Explanation of symbols]

[0094] 10...Optical sorting device 20...Detection unit 30...First light source 31...Second light source 32...First Light 33...Second Light 40...Area sensor 41...Angle of view 60...Sorting Department 61...Nozzle 62,62a~62i...Aperture 63...Valve 64...Air 65...Sorting area 71...Storage tank 72...Feeder 73...Shoot 74...Good product discharge trough 75...Defective product discharge trough 80...Controller 81...Identification section 82... Arithmetic section 83...Injection control unit 84...Monitoring Department 85...Prediction Section 86...Identification processing section 90...Object 90,90a~90i,91,92,93,94...Target 93a,94a...center of gravity 93b,94b...Long axis 95,95a,95d,95g,96,97,97g...vector information 110...Sorting device 140...First area sensor 141...angle of view 145...Second area sensor 146...angle of view D1...transfer direction D2...Chute width direction

Claims

1. An inspection device for inspecting a condition of an object, comprising: A transport unit configured to transport the object; an area sensor configured to capture an image of the object being transported by the action of the transport unit at a frame rate capable of capturing an image of the same object multiple times; an identification unit configured to identify the state of the object based on an image acquired by the area sensor; An inspection device comprising:

2. An inspection device for inspecting a condition of an object, comprising: A transport unit configured to transport the object; an area sensor configured to capture an image of the object being transported by the action of the transport unit at a frame rate capable of capturing an image of the same object multiple times; an identification unit configured to identify the state of the object based on an image acquired by the area sensor; a calculation unit configured to calculate behavior information of the same object during the multiple imaging operations based on multiple images obtained by imaging the same object multiple times; Equipped with The behavior information includes at least one of size change information related to a change in size of the same object between the multiple images, and vector information representing at least one of a change in position and a change in posture of the same object. Inspection equipment.

3. A sorting device comprising: The inspection device according to claim 2 ; a sorting unit configured to selectively inject air toward a specific object determined based on an identification result of the identification unit in accordance with a timing at which the specific object passes through a sorting region, thereby changing a transfer trajectory of the specific object, thereby sorting the specific object; an injection control unit configured to determine an injection mode of the air based on at least one of the size change information and the vector information; A sorting device comprising:

4. The sorting device according to claim 3, the area sensor has an angle of view capable of capturing an image of both an object located upstream of the sorting area and an object located downstream of the sorting area on a transport path of the object, and is configured to obtain an upstream image in which the object located upstream of the sorting area is captured, and a downstream image in which the object located downstream of the sorting area is captured, The identification unit is configured to identify the state of the object based on the upstream image. Sorting device.

5. The sorting device according to claim 4, a monitoring unit configured to monitor an operation status of the air injection by the sorting unit based on the downstream side image. Sorting device.

6. A sorting device for sorting objects, comprising: A transport unit configured to transport the object; an area sensor configured to capture an image of the object being transported by the action of the transport unit at a frame rate capable of capturing an image of the same object multiple times; an identification unit configured to identify the state of the object based on an image acquired by the area sensor; a sorting unit configured to selectively inject air toward a specific object determined based on an identification result of the identification unit in accordance with a timing at which the specific object passes through a sorting region, thereby changing a transfer trajectory of the specific object, thereby sorting the specific object; Equipped with The area sensor has an angle of view capable of capturing an image of the object both upstream of the sorting area and downstream of the sorting area on the transport path of the object. Sorting device.

Images