JP6950994B1 - Fish fillet bone removal system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a fish fillet bone extraction system capable of recognizing the position of each bone with respect to a fish fillet and extracting only the bone without damaging the fish body around the bone. SOLUTION: In the fish fillet bone extraction system of the present invention, a fish fillet 21 is placed on a pallet 20, and by moving the pallet 20, bone is extracted from the fish fillet 21 placed on the pallet 20 and placed on the pallet 20. In the first imaging step 3A for imaging the placed fish feeler 21, the bone region specifying step 13A for identifying the region where the bone exists from the first imaging image captured in the first imaging step 3A, and the bone region specifying step 13A. It was identified in the second imaging step 3B for imaging the specified region, the bone position specifying step 13B for identifying the bone position by the second imaging image captured in the second imaging step 3B, and the bone position specifying step 13B. It is characterized by having a bone removal step 3C in which the outer end portion of the bone is grasped by the bone removal chuck 42 and the bone is removed by moving the bone removal chuck 42. [Selection diagram] Fig. 1

Description

The present invention relates to a fish fillet bone extraction system for accurately extracting small bones and pin bones from fish meat, particularly the half body of fish cut into three pieces.

Currently, in Japan, the salmon market is in demand of over 400,000 tons per year. Most of the salmon and salmon production in Japan is from natural catches, which generally matches the tastes of Japanese people who like "natural products", but on the other hand, there are problems such as parasites and "raw food". Not suitable for. Due to this industrial structure, imports are increasing, and considering that "raw food" processed products generally have higher added value than "heating" processed products, it is profitable for Japanese producers and processing companies as a result. It is a situation where the added value as a product is hard to drop.
In Japan, salmon and salmon are unsuitable for "raw food" because salmon and salmon are one of the most established fish species in Japan, and it is difficult to increase the natural catch due to resource constraints. Aquaculture is expanding, and some shifts from imported products to domestic products are expected.
The salmon processing process consists of scale removal, head cutting, filleting, skinner, and pinbone removal. Currently, scale removal, headcut, fillet, and skinner can be automated regardless of salmon size using existing fish processing machines. However, the work of deboning salmon's superior nerve bone (small bone or pinbone) still relies on human labor, which requires a great deal of labor. A device for automatically extracting pinbones has also been devised, but there are various problems. It will be explained according to the reference.
Patent Document 1 is a device for pulling out a pinbone of a fish fillet by sandwiching it between a set of belts that rotate so as to be involved. However, since there are individual differences in the shape of the fish and the position where the bone is located, if the bone is misaligned with respect to the belt, the boneless portion may be sandwiched between the belts and the bone may be left behind.
Patent Document 2 is a mechanism for pinching and pulling out a pinbone protruding from a fish fillet with a specific holding member. Therefore, when the position of the holding member and the pinbone are displaced due to buried bones or individual differences in the fish fillet. Does not come out well or is left behind.
In Patent Document 3, a rotating roller is pressed against a fish fillet to rotate the rotating roller in a direction opposite to the traveling direction of the fish fillet, and the roller is made of a net-like structure through which bones pass. The pinbone is pulled out by passing the bone through the net, but the pinbone needs to protrude from the fillet.
Patent Document 4 has a structure in which a pinbone is projected by pressing a roller against a fish feeler, and the pinbone is pulled out by a jig such as a claw while maintaining the state. Since there are individual differences, if the bone is misaligned with respect to the jig position, the bone may be left behind.
As described above, these devices are limited to the bones that are pierced perpendicular to the surface of the fish body, or the fish body is pressed by the bone removal device to project the pin bones. There is a problem of transforming. Furthermore, a common problem with these devices is that the bone is not removed after the bone is located, but the bone is removed at a position where the bone is likely to be left behind. In some cases, the fish feel may be discarded because it may be injured by touching or touching the surrounding body. For this reason, salmon watering is mainly done manually, and due to these problems, automation has not been possible at present.
In Patent Document 5, the position of the bone is confirmed by an image, but there are variations in the size and shape of the fish fillet, the position and direction of the bone, and so on. It is difficult, and there is a problem that it is not possible to measure the position of the bone three-dimensionally, recognize the direction in which the bone is stuck by the measurement, and extract the bone by three-dimensional movement along the direction. Therefore, there are cases where the bones are forcibly pulled out and the body is hurt or left behind, but the recovery against it is not sufficient and it is not in the form of a fully automatic system.

Japanese Patent No. 4955036 Japanese Unexamined Patent Publication No. 2001-61404 Special Fair 07-063300 Japanese Patent No. 40782281 Japanese Patent No. 600568

Patent Documents 1 to 4 have been proposed as a device for automatically removing the bone of a fish fillet as in the present invention. However, these devices are limited to the bones that are pierced perpendicular to the surface of the fish body, and the body is used to hold the fish body in order to bring the fish body into contact with the water removal device, or to lift the fish body from below. There is a problem of transforming. Furthermore, a common problem with these devices is that the bone is not removed after the bone is located, but the bone is removed at a position where the bone is likely to be left behind. In some cases, the fish feel may be discarded because it may be injured by touching or touching the surrounding body.
For this reason, salmon watering is mainly done manually, and due to these problems, automation has not been possible at present.
In recent years, in the backyards of major cities, it is difficult to secure employees, and there is also a shortage of skilled workers, so higher-order processing including watering work is required for production areas. However, securing skilled workers in production areas is becoming more difficult year by year with the aging of workers, and there is a demand for mechanization and unmanned production area processing plants. Furthermore, the Japanese food boom is accelerating all over the world, and the demand for frozen sashimi cuts and frozen sushi ingredients is increasing year by year.
However, instead of pulling out only the bone so as not to hurt the fish, it is possible to touch the body around the bone and injure the body in some cases. Furthermore, in response to the recent increase in demand for boneless fillets, watering of small fish such as horse mackerel and Thailand is also required, and a fully automatic bone-removing system corresponding to various fish types and sizes is required. There is still no system that can solve such problems.

An object of the present invention is to provide a fish fillet bone extraction system capable of recognizing the position of each bone with respect to a fish fillet and extracting only the bone without damaging the fish body around the bone. ..

The fish fillet bone removal system of the present invention according to claim 1 is a fish fillet bone removal system in which a fish fillet is placed on a pallet and bone is removed from the fish fillet placed on the pallet by moving the pallet. Therefore, a first imaging step of imaging the fish fillet placed on the pallet, a bone region specifying step of identifying a region where the bone exists from the first imaging image captured in the first imaging step, and a bone region specifying step. A second imaging step of imaging the region specified in the bone region specifying step, a bone position specifying step of identifying the bone position by the second imaging image captured in the second imaging step, and the bone position. It is characterized by having a bone removal step of grasping the outer end portion of the bone specified in the specific step with a bone removal chuck and removing the bone by moving the bone removal chuck.
The present invention according to claim 2 is characterized in that, in the fish fillet bone extraction system according to claim 1, the region in the bone region specifying step is specified by machine learning.
According to the third aspect of the present invention, in the fish fillet bone removal system according to claim 1 or 2, after the bone removal step, an inspection step of inspecting whether or not the bone remains in the fish fillet is performed. , Characterized by having.
The present invention according to claim 4 is characterized in that, in the fish fillet bone extraction system according to claim 3, an X-ray image is used in the inspection step.
The present invention according to claim 5 is characterized in that, in the fish fillet bone extraction system according to claim 3, an infrared image is used in the inspection step.
The present invention according to claim 6 is described in the case where it is determined in the inspection step that the bone does not remain in the fish fillet bone extraction system according to any one of claims 3 to 5. When the pallet on which the fish fillet is placed is moved to a non-defective product storage location and it is determined in the inspection process that the bone remains, a defect coefficient indicating the number of defects is added and the defect coefficient is a predetermined number of times. When the above is not reached, the pallet on which the fish fillet is placed is moved to the bone removal step again, and even when it is determined in the inspection step that the bone remains, the defect coefficient is the predetermined value. When the number of times has been reached, the pallet on which the fish fillet is placed is moved to a defective product storage location.
The present invention according to claim 7 is for measuring at least one of the position and inclination of the pallet on the pallet in the fish fillet bone extraction system according to any one of claims 1 to 6. 3 or more reference marks and a marker that can identify the pallet are attached, and one of the fish fillets is placed on the pallet so as not to hide the reference mark, and the pallet on which the fish fillet is placed is placed. Is moved to the first imaging step, the second imaging step, and the watering step, and in the first imaging step, the second imaging step, and the watering step, the coordinate correction is performed by reading the reference mark. It is characterized by doing.
According to the eighth aspect of the present invention, in the fish fillet bone extraction system according to the sixth aspect, the pallet is provided with three or more reference marks for measuring at least one of the position and inclination of the pallet. A marker that can identify the pallet is attached, and one piece of the fish fillet is placed on the pallet so as not to hide the reference mark, and the pallet on which the fish fillet is placed is subjected to the first imaging step. The process is moved to the second imaging step, the watering step, and the inspection step, and in the first imaging step, the second imaging step, the watering step, and the inspection step, coordinate correction is performed by reading the reference mark. In the defective product work place for moving the pallet in the defective product storage location, a device for reading the marker and a display for displaying the inspection result of the fish fillet placed on the pallet specified by the marker. It is characterized by having.

According to the fish fillet bone extraction system of the present invention, the position of each bone with respect to the fish fillet can be recognized one by one, and only the bone can be extracted without damaging the fish body around the bone.

A process diagram showing a fish fillet bone extraction system according to Example 1 of the present invention. Configuration diagram showing the pallet used in the fish fillet bone removal system Configuration diagram showing the work location of the fish fillet bone removal system A block diagram showing a working place of the fish fillet bone extraction system according to the second embodiment of the present invention. A block diagram showing a working place of the fish fillet bone extraction system according to the third embodiment of the present invention.

The fish fillet bone extraction system of the present invention identifies a bone region in which a region where bone exists is specified by a first imaging step of imaging a fish fillet placed on a pallet and a first imaging image captured in the first imaging step. A step, a second imaging step of imaging the region specified in the bone region specifying step, a bone position specifying step of identifying the bone position by the second imaging image captured in the second imaging step, and a bone position specifying step. It has a bone removal process in which the outer end of the bone specified in is grasped with a bone removal chuck and the bone is removed by moving the bone removal chuck. Only the bone can be extracted without damaging the fish meat.

FIG. 1 is a process diagram showing a fish fillet bone extraction system according to the first embodiment of the present invention.
At the loading location 1, the fish feeler 21 (see FIG. 2) is placed on the pallet 20 (see FIG. 2). The pallet 20 on which the fish feel 21 is placed at the loading location 1 is moved by the conveyor 11 and moved to the pallet merging location 2.
At the loading location 1, the marker 22 (see FIG. 2) on the pallet 20 is read and sent to the system computer 13 via the communication line 12. As a result, the marker 22 information of the palette 20 and the fish feel 21 are specified. The reading device and reading method of the marker 22 are suitable for non-contact magnetic information reading, but other methods may be used. When the pallet 20 moves from the loading location 1 to the merging location 2, the new fish feel 21 is placed on the new pallet 20 at the loading location 1 almost at the same time.
The pallet 20 not equipped with the fish feel 21 is preferably automatically supplied from a stocker capable of storing a plurality of pallets 20, but may be supplied by a person.
The pallet merging place 2 is also a retry loading place, and the pallet 20 for retrying from the retry conveyor 10 is controlled so as not to collide with the pallet 20 from the loading place 1. When the pallet 20 for retry exists on the retry conveyor 10, the pallet 20 from the charging place 1 and the pallet 20 from the retry conveyor 10 are alternately supplied to the merging place 2 and moved to the work place 3. ..

The work place 3 has a first imaging step 3A, a second imaging step 3B, and a watering step 3C.
In the first imaging step 3A, the fish feeler 21 placed on the pallet 20 is imaged, and the system computer 13 identifies the region where the bone exists from the first captured image captured in the first imaging step 3A (bone region). Specific step 13A).
In the second imaging step 3B, the region specified in the bone region specifying step 13A is imaged, and in the system computer 13, the bone position is specified by the second captured image captured in the second imaging step 3B (bone position identification). Step 13B).
In the bone removal step 3C, the outer end of the bone specified in the bone position specifying step 13B is grasped by the bone removal chuck 42, and the bone is removed by moving the bone removal chuck 42.
At the work place 3, the bones of the fish fillet 21 mounted on the pallet 20 are pulled out one by one in this way. Details will be described later with reference to FIGS. 2 and 2.

The pallet 20 after the watering step 3C is moved to the inspection site 4. At the inspection place 4, it is judged by an image whether or not there is a bone residue left by X-ray or infrared rays (inspection step). The determination result is sent to the system computer 13 via the communication line 12 together with the information of the marker 22 read at the inspection place 4, and the information of the marker 22 and the watering result are specified. The determination result includes at least the number of times the same fish feel 21 is determined to be defective and the location information of the remaining bone. The marker 22 is read at the non-defective product branching location 5, and if all the predetermined bones have been removed, the marker 22 is moved from the non-defective product branching location 5 to the non-defective product storage location 7. The non-defective product storage location 7 is a group of non-defective product stockers that store the pallet 20 on which the fish fillet 21 determined to be non-defective product is mounted. It is also possible to make a system that always takes out without storing. The marker 22 is read at the non-defective product branching location 5, and if it is defective, the marker 22 is moved to the defective retry branching location 6. That is, the number of times the fish feeler 21 on which the pallet 20 is mounted is determined to be defective by the marker 22 read at the defective retry branching location 6 is compared with a predetermined number of times to reach a predetermined number of times. If so, move to the defective product storage location 8. The defective product storage location 8 is a defective product stocker that stores the pallet 20 on which the fish fillet 21 determined to be defective is mounted. It is possible to make a system in which the system is always taken out without storing it, or it is possible to make a system in which the worker 17 moves to the defective work place 16. If the number of defects of the fish fillet 21 mounted on the pallet 20 has not reached the predetermined number of times by the marker 22 read at the defective retry branching location 6, the fish fillet 21 is moved to the retry conveyor 10 and sent to the merging location 2. Be done.

In this embodiment, the defective pallet group on which the fish fillet 21 with leftover bone is placed is shown as the defective product storage location 8. The defective product stocker at the defective product storage location 8 is carried to the defective product stocker storage location 15. The defective product work place 16 is a place where one pallet 20 is taken out from the defective product stocker carried to the defective product stocker storage place 15 and placed on the defective product stocker. Here, the marker 22 of the pallet 20 is read and the defective product information is displayed on the display 14. do. The worker 17 removes the leftover bone according to the display content of the display 14. The defective product storage location 8 may be used as the defective product work location 16.

FIG. 2 is a block diagram of a pallet used in the fish fillet bone extraction system according to the present embodiment.
FIG. 2 shows a state in which the fish fillet 21 is placed on the pallet 20. It is preferable to provide three or more reference marks 23, and at least one of the position and inclination of the pallet 20 is measured. Although not shown, a measurement sensor capable of measuring the coordinates of the reference mark 23 is arranged above the pallet 20, and the coordinates of the reference mark 23 are measured by the measurement sensor to specify the coordinates of the pallet 20 with respect to the position of the measurement sensor. be able to. Even when the pallet 20 moves to another work place 3, the position of the pallet 20 is specified by reading the reference mark 23 of the moved pallet 20 by the measurement sensor in the same manner. Therefore, even if the stop position of the pallet 20 is deviated, the shape of the fish fillet 21 placed on the pallet 20 can be changed by reading the reference mark 23 at all work places where the reference mark 23 is located. The shape can be shared in all work areas. If the stop position of the pallet 20 does not shift or is within the permissible range at all work locations, it is not necessary to correct the coordinates using this reference mark 23. Therefore, if it is possible to realize an environment in which the work is performed in such a way that the stop position of the pallet 20 does not shift, it is possible to eliminate the reference mark 23.

The marker 22 specifies the pallet 20, and the fish feel 21 on the pallet 20 can also be specified. In this embodiment, when the pallet 20 is carried into the work place 3, the marker 22 is also read almost at the same time as the reference mark 23, and the content of the read marker 22 is the system computer on the communication line 12 by LAN or wireless LAN. Sent to 13. At the inspection location 4, the inspection result corresponding to the marker 22 on the pallet 20 is sent to the system computer 13. At the non-defective product branch location 5, the system computer 13 determines from the marker 22 information whether it should be sent to the defective retry branch location 6 or the non-defective product storage location 7. The same applies to the defective work place 16, and the system computer 13 displays the inspection result corresponding to the marker 22 on the display 14.

FIG. 3 is a configuration diagram showing a working place of the fish fillet bone extraction system according to the present embodiment.
In the first imaging step 3A, when the pallet 20 is carried in by the conveyor 11, the reference mark 23 and the marker 22 are read, and the information is transferred to the system computer 13. The three-dimensional sensor 36 measures the height information and the brightness information in a line shape, and captures the three-dimensional coordinates of the entire fish feel 21. If you do not want to read the height information, you can use a 2D camera. The three-dimensional sensor 36 is mounted on the rail 44 and measures while moving along the rail 44. In this embodiment, the conveyor 11 moves in the direction facing the moving direction 37. When the measurement of the entire fish feeler 21 or the determined area is completed, the three-dimensional information is transferred to the system computer 13 to identify a likely part or subgroup of the bone of the fish feeler 21. An image processing computer different from the system computer 13 may be used for identification by image processing, and the specified coordinates may be transferred to the system computer 13. Further, in the image processing computer, the system computer 13, or a computer attached thereto, processing can be performed by a machine-learned neural network instead of image processing. In that case, it is possible to improve the recognition accuracy by accumulating the captured image in the memory and further machine learning the accumulated image. Then, in the memory managed by the system computer 13 together with the reference mark 23, the marker 22 information, and the like, the probable parts or subgroups of the bones of the fish feel 21 are stored as coordinates representing those areas. Almost at the same time as the completion of the three-dimensional measurement, the pallet 20 on which the fish feeler 21 is mounted is carried into the second imaging step 3B.

In the second imaging step 3B, the reference mark 23 and the marker 22 are read and the information is transferred to the system computer 13 as in the first imaging step 3A. The articulated robot 38 is equipped with a lighting 39 and a camera 40 at its tip. The system computer 13 transfers to the articulated robot 38 the coordinates of the likely bone or subgroup of the fish feeler 21 calibrated by the reference mark 23, which corresponds to the marker 22 information. The illumination 39 and the camera 40 mounted on the articulated robot 38 move to a place where bones are likely to be transferred according to the transferred information, and perform imaging.
In the second imaging step 3B, the image information is transferred to the image processing computer, the bone or bone group of the fish fillet 21 is specified, and the specified coordinates are transferred to the system computer 13. When a two-dimensional camera is used instead of the three-dimensional sensor 36 in the first imaging step 3A, a camera that can simultaneously obtain height information is used instead of the camera 40, or a height sensor is used separately from the camera 40. In addition, it is transferred to the system computer 13 as three-dimensional coordinates. As in the case of the first imaging step 3A, image information may be transferred to the system computer 13 for image processing due to capital investment and production tact, or a machine-learned neural network may be used instead of image processing. It can also be processed. In that case as well, as in the case of the first imaging step 3A, it is possible to improve the recognition accuracy by accumulating the captured images in the memory and further machine learning the accumulated images. Then, the bones or bone groups of the fish feel 21 are stored in the memory managed by the system computer 13 together with the reference mark 23 and the marker 22 information. The recognized bone is stored in a memory managed by the system computer 13 with the location of the bone sandwiched by the bone removal chuck 42 for removing the bone as coordinates. In addition to the coordinates, the direction in which the bone is stuck may also be stored in the memory. The direction of bone removal is basically the direction of the fish's head, but if the direction is different due to individual differences, or if the fish body and bone are strongly entwined and it is difficult to remove the bone, follow the direction in which the bone is stuck. This is because it may be easier to pull it out.

Almost at the same time as the end of the image measurement, the pallet 20 on which the fish feel 21 is mounted is carried into the watering step 3C.
Similar to the first imaging step 3A and the second imaging step 3B, the reference mark 23 and the marker 22 are read, and the information is transferred to the system computer 13. The articulated robot 41 is equipped with a watering chuck 42 at its tip. The system computer 13 transfers the coordinates of the bone portion or subgroup of the fish feeler 21 calibrated by the reference mark 23, which corresponds to the marker 22 information, to the articulated robot 41. The bone removal chuck 42 moves to the part with the bone according to the transferred information and removes the bone. This bone removal step 3C is repeated for a predetermined number of bones. The result is transferred to the system computer 13, and after the boning step 3C is completed, the pallet 20 is moved to the inspection step.

The reason why the articulated robot 38 is used in the second imaging step 3B is that the fish fillet 21 has a three-dimensional shape. High measurement is possible. For example, by measuring at a position perpendicular to the surface of the fish feeler 21, image distortion and the like are reduced at the same time that the entire surface is in focus. The reason why the articulated robot 41 is used in the watering step 3C is that the bone stuck in the fish fillet 21 basically faces the direction of the fish's head, but there are variations. Since the bones are entwined with the body of the fish, they can be pulled out smoothly by pulling them out in the direction in which the bones are facing. The bone removal chuck 42 is mounted on the articulated robot 41 to remove the bone in various directions, and the outer end of the bone is grasped by the bone removal chuck 42.

As a further advantage of attaching the camera 40, the image of the camera 40 is used to focus by a mechanism that moves the camera 40 back and forth. The distance to the bone removal chuck 42 becomes constant, so that the distance to the bone removal chuck 42 can be known, and the bone removal chuck 42 can be accurately guided to the position of the bone. Instead of focusing and measuring the distance with the camera 40, it is possible to change the camera 40 to a three-dimensional sensor 36 or add a height sensor separately from the camera 40 to take three-dimensional coordinates. ..

Although not shown in the figure, the reference mark 23 and the marker 22 are read and the information is transferred to the system computer 13 also in the inspection process. Then, the bone left unextracted is determined from the image by X-ray or infrared ray. The acquired image is transferred to an image processing computer (not shown), determines whether or not there is any bone left behind by the fish fillet 21, and transfers the result to the system computer 13. The image may be transferred to the system computer 13 instead of the image processing computer for processing due to capital investment and production tact. Instead of image processing, it can be processed by a machine-learned neural network. In that case, it is possible to improve the recognition accuracy by accumulating the captured image in the memory and further machine learning the accumulated image. If it is determined from the judgment result that there is leftover bone, the number of inspections for the same fish feeler 21 is performed by inquiring the read marker 22 to the system computer 13 via the communication line 12. And increase the number of inspections (hereinafter referred to as the defect coefficient) by one. The defect coefficient is set to 0 when the fish feeler 21 is newly added to the pallet 20. The coordinates of the location of the leftover bone are transferred to the system computer 13 at the coordinates calibrated by the reference mark 23. Instead of transferring the coordinates calibrated by the reference mark 23 to the system computer 13, the coordinates of the leftover bone may be transferred to the system computer 13 and converted into the coordinates calibrated by the reference mark 23 by the system computer 13. .. When transferring the coordinates to the system computer 13, the number of bones left behind is also transferred.

When the defect coefficient of the pallet 20 carried into the work place 3 shown in FIG. 1 is 1 or more by reading the marker 22, the pallet 20 passes the first imaging step 3A and the second imaging step 3B, and the bone removal step 3C. It is also possible to perform bone removal based on the calibrated coordinates of the leftover bone from the system computer 13. At this time, if it is determined that the fish feeler 21 mounted on the pallet 20 may have moved with respect to the pallet 20 during the retry conveyor 10 or other transportation, the first imaging step 3A and the second imaging step 3B are performed. The work may be performed without passing, or the work may be performed from the second imaging step 3B after passing the first imaging step 3A. In this case, instead of using the coordinates of the leftover bone stored in the system computer 13, the bones are removed using the coordinates of the bone newly measured in the first imaging step 3A and the second imaging step 3B.

FIG. 4 is a configuration diagram showing a working place of the fish fillet bone extraction system according to the second embodiment of the present invention.
In this embodiment, in order to reduce the number of steps from the balance of investment amount, work tact, production amount, etc., the first imaging step 3A remains the same as that of the first embodiment, and the second imaging step 3B and the watering step 3C are the same. It is done at the place. In step 3D, the second imaging step 3B and the watering step 3C are performed by the same articulated robot 51. That is, the lighting 39, the camera 40, and the watering chuck 42 are mounted on one articulated robot 51.

The new advantage of this embodiment is that after the bone is removed, it is confirmed by the lighting 39 and the camera 40, and when it is determined that the bone is not removed, the bone can be removed again. As a method of determining that the bone is not missing, there is a method of capturing an image in the vicinity of the place where the bone should be, performing image processing, and confirming the bone, but as another method, the bone removal chuck 42 is in the field of view of the camera 40. There is also a method of confirming the bone by taking an image around the tip of the bone removal chuck 42 and performing image processing after removing the bone by devising so that the area around the tip of the bone is reflected. That is, when there is bone around the tip of the bone removal chuck 42, it can be seen that the bone is missing. For that purpose, if the previously removed bone remains attached to the vicinity of the tip of the bone removal chuck 42, a correct judgment cannot be made. Therefore, it is necessary to remove the bone from the bone removal chuck 42 every time after removing the bone. The removal method can be performed by removing the bone from the tip portion of the bone removal chuck 42 and then washing the bone. After further washing, it is possible to confirm the presence of bone on the image. Further, in the first embodiment, when the pallet 20 moves from the second imaging step 3B to the bone removal step 3C, if the fish fillet 21 moves for some reason, the coordinates of the bone sent to the articulated robot 41 in the bone removal step 3C are used. There may be a discrepancy that makes it difficult to remove the bone, but as explained earlier, this problem can be solved by recognizing the bone again and controlling it to remove the bone. Further, in the second embodiment, it is possible to control the bone removal chuck 42 while checking the position of the bone to remove the bone, and it is possible to realize a bone removal system with improved reliability. Further, when the confirmation of the bone residue of the fish fillet 21 after the bone removal is performed in Example 2 so as to be performed every time all the bones are removed, the inspection step can be omitted.

As in the second embodiment, as a further advantage of attaching the camera 40, the image of the camera 40 is used to focus by a mechanism that moves the camera 40 back and forth. The distance to the image, for example, the bone to be removed, becomes constant, so that the distance to the bone removal chuck 42 can be known, and the bone removal chuck 42 can be accurately guided to the position of the bone. A sensor that takes height information can be used instead of the camera 40, or a height sensor can be added separately from the camera 40 to take three-dimensional coordinates.

FIG. 5 is a block diagram showing a working place of the fish fillet bone extraction system according to the third embodiment of the present invention.
In the third embodiment, the first imaging step 3A is the same as that of the first embodiment, and the second imaging step 3B and the watering step 3C are performed at the same place, but in the step 3E, the second imaging step 3B is performed by an articulated robot. 38 is performed, and the watering step 3C is performed by the articulated robot 41. That is, the lighting 39 and the camera 40 are mounted on the articulated robot 38, and the watering chuck 42 is mounted on the articulated robot 41.

The advantages that Example 3 has newly added to Example 2 are as follows. The bone removal chuck 42 and the removed bone can be freely confirmed by the illumination 39 and the camera 40. For example, in Example 2, depending on the position of the camera 40 and the position of the bone removal chuck 42, even if only half of the removed bone is shown, the camera 40 can be freely moved, so that the entire bone can be photographed, and the bone can be photographed. You can check the overall shape.

As a further advantage, in the second embodiment, the illumination 39 and the camera 40 move to a place where the bone-containing place can be photographed, and after the photographing, the bone-removing chuck 42 moves to the bone position. In Example 3, when the illumination 39 and the camera 40 take a picture in a place containing bone, the bone removal chuck 42 can be moved to the vicinity of the bone almost at the same time, so that the bone is removed after recognizing the bone. It is possible to reduce the moving time of the articulated robot 38 up to. In the third embodiment, two articulated robots 38 and 41 are used, but this can be realized by a dual-arm robot having one robot and two arms. As in the first and second embodiments, a sensor that takes height information can be used instead of the camera 40, or a height sensor can be added separately from the camera 40 to take three-dimensional coordinates.

In the present invention, the pinbone is accurately extracted from the half body of the fish cut into three pieces.

1 Input place 2 Pallet merging place 3 Work place 4 Inspection place 5 Good product branching place 6 Defective retry branching place 7 Good product storage place 8 Defective product storage place 10 Conveyor 11 Retry conveyor 12 Communication line
13 System computer 14 Display 15 Defective product stocker storage location 16 Defective product work location 17 Worker 20 Pallet 21 Fish feel 22 Marker 23 Reference mark 36 3D sensor 37 Movement direction 38 Articulated robot 39 Lighting 40 Camera 41 Articulated robot 42 Watering Chuck 44 Rail 51 Articulated robot

Claims (8)

A fish fillet bone extraction system in which a fish fillet is placed on a pallet and bones are removed from the fish fillet placed on the pallet by moving the pallet.
The first imaging step of imaging the fish fillet placed on the pallet, and
A bone region specifying step of identifying a region where the bone exists from the first captured image captured in the first imaging step, and a bone region specifying step.
A second imaging step of imaging the region specified in the bone region specifying step, and
A bone position specifying step of specifying the position of the bone by the second captured image captured in the second imaging step, and a bone position specifying step.
A fish fillet bone removal system comprising a bone removal step of grasping the outer end portion of the bone specified in the bone position specifying step with a bone removal chuck and removing the bone by moving the bone removal chuck. The fish fillet bone extraction system according to claim 1, wherein the region in the bone region specifying step is specified by machine learning. After the watering step, an inspection step of inspecting whether or not the bone remains in the fish fillet is performed.
The fish fillet bone extraction system according to claim 1 or 2, wherein the fish fillet is provided. The fish fillet bone extraction system according to claim 3, wherein an X-ray image is used in the inspection step. The fish fillet bone removal system according to claim 3, wherein an infrared image is used in the inspection step. If it is determined in the inspection step that the bone does not remain, the pallet on which the fish fillet is placed is moved to a non-defective product storage location.
When it is determined in the inspection step that the bone remains, a defect coefficient indicating the number of defects is added, and when the defect coefficient does not reach the predetermined number, the pallet on which the fish fillet is placed. Is moved to the watering step again,
Even when it is determined in the inspection step that the bone remains, when the defect coefficient reaches the predetermined number of times, the pallet on which the fish fillet is placed is moved to the defective product storage location. The fish fillet bone extraction system according to any one of claims 3 to 5, characterized in that. The pallet has
Three or more reference marks for measuring at least one of the positions and tilts of the pallet,
A marker that can identify the palette is attached,
On the pallet, one piece of the fish fillet is placed so as not to hide the reference mark.
The pallet on which the fish fillet is placed is moved to the first imaging step, the second imaging step, and the watering step.
The fish according to any one of claims 1 to 6, wherein in the first imaging step, the second imaging step, and the watering step, coordinate correction is performed by reading the reference mark. Fillet bone removal system. The pallet has
Three or more reference marks for measuring at least one of the positions and tilts of the pallet,
A marker that can identify the palette is attached,
On the pallet, one piece of the fish fillet is placed so as not to hide the reference mark.
The pallet on which the fish fillet is placed is moved to the first imaging step, the second imaging step, the watering step, and the inspection step.
In the first imaging step, the second imaging step, the watering step, and the inspection step, coordinate correction is performed by reading the reference mark.
In the defective product work place where the pallet is moved in the defective product storage location,
A device that reads the marker and
The fish fillet bone removal system according to claim 6, further comprising a display for displaying an inspection result of the fish fillet placed on the pallet specified by the marker.

Images