Medicine box medicine taking method and medicine box storing and taking device based on image model matching recognition

Abstract

The invention relates to a medicine box medicine taking method and a medicine box storing and taking device based on image model matching identification, the medicine box medicine taking method and the medicine box storing and taking device based on the image model matching identification are ingenious in design, key feature values on multiple faces of a medicine box are extracted and learned to be bound with the medicine box, and the key feature values are uploaded to a medicine box database; then the medicine boxes with high similarity are placed in different trays according to different specifications, the recognition error can be reduced, the medicine boxes are shot through the first camera, and the medicine boxes in the trays are determined through comparison recognition according to the corresponding medicine box information in the database and the key feature values of the shot medicine box images; after the tray is put on a shelf, the two-dimensional coordinate information of the medicine boxes on the tray is shot by the second camera and is converted into the three-dimensional coordinate information of the medicine boxes through space calculation in combination with the height information of the medicine boxes, and then according to the three-dimensional coordinate of the target medicine box, the line with the shortest goods taking path is found, so that the medicine can be taken. And the accuracy is high.

Description

Technical Field

[0001] The present invention relates to the technical field of medicine box recognition, and more specifically to a medicine box taking method and a medicine box storage and access device based on image model matching and recognition. Background Art

[0002] Most existing image recognition methods are based on convolutional neural network recognition. For some medicine boxes with very high similarity in outer packaging, such as medicine boxes of the same brand and type but different specifications, this method based on convolutional neural network recognition has uncertainty and its recognition accuracy can only reach about 98%. For medicine boxes, even a 2% error is unacceptable. Therefore, a recognition method that can accurately identify medicine boxes without error is needed. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a medicine box taking method and a medicine box storage and retrieval device based on image model matching and recognition in view of the above-mentioned defects of the prior art.

[0004] On the one hand, the technical solution adopted by the present invention to solve its technical problem is: a method for taking medicine from a medicine box based on image model matching and recognition, which includes the following steps:

[0005] In the medicine box identification area, a first camera is used to collect one or more faces of the medicine box with obvious features and learn key feature values ​​on each face, and then the name and model information of the medicine box are bound by the key feature values ​​to establish a feature database;

[0006] After entering the information of the medicine boxes that need to be replenished, the medicine boxes are classified and managed according to multiple key feature values ​​of the medicine boxes, and then the medicine boxes of the same brand and type are placed in different trays according to different specifications;

[0007] Take a photo of the medicine boxes in the tray again to determine the type of medicine boxes, and then compare and identify the corresponding medicine box information in the database with the key feature values ​​of the photographed medicine box image to determine the medicine boxes in the tray again. If the comparison results are the same, the medicine boxes are put on the shelf. If the comparison results are different, they are returned to the waiting area.

[0008] After the pallet is put on the shelf, the second camera is controlled to take pictures of the pallet and the medicine boxes on the pallet, and the coordinates of the feature points of the calibrated image are superimposed and overlapped with the coordinates of the actual pallet to obtain the two-dimensional coordinates of all the medicine boxes in the pallet in the image; after the two-dimensional plane coordinates of the medicine boxes in the image are converted into three-dimensional space coordinates, the shortest route is found according to the three-dimensional coordinates of the target medicine box to pick up the medicine;

[0009] In the medicine box taking method based on image model matching and recognition of the present invention, the second camera cooperates with one or more fill lights around it to continuously shoot multiple images with different exposures of the same surface of the medicine box on the tray;

[0010] Preprocess the collected multiple images with different exposures, and transform the multiple images with different exposures into the same coordinate system through image registration algorithm;

[0011] The registered multiple images are fused through a fusion algorithm to obtain image data of the current surface feedback; the image data is then compared with the information of the medicine box in the database to determine the target medicine box;

[0012] In the medicine box taking method based on image model matching and recognition of the present invention, the steps of converting the two-dimensional plane coordinates of the medicine box on the image into three-dimensional space coordinates are as follows:

[0013] The height of the medicine box is identified by the distance measuring sensor, and then the height of the medicine box and the two-dimensional plane coordinates of the medicine box on the image are spatially calculated to obtain the three-dimensional coordinates of the current medicine box;

[0014] In the medicine box taking method based on image model matching and recognition described in the present invention, the distance measuring sensor can be set in the recognition area, and the height of the medicine box and related key feature values ​​are bound and stored in the feature database in advance.

[0015] The medicine box taking method based on image model matching and recognition described in the present invention, wherein deformation calculation is performed based on the second camera height parameter, focal length parameter, viewing angle parameter and the laser coefficient of the second camera itself, the internal parameters and distortion parameters of the second camera are obtained, and the image deformation law is generated; after the second camera takes a panoramic photo of the medicine boxes in the tray, the panoramic image taken by the camera is corrected according to the image deformation law to obtain the actual positions of all medicine boxes in the image.

[0016] The medicine box retrieval method based on image model matching and recognition described in the present invention is characterized in that the tray is provided with a unique identification code; before the medicine box is placed in the tray, the identification code of the tray is obtained and the medicine box information is associated with the identification code to generate a list of the medicine boxes in the current tray.

[0017] The method for taking medicine from a medicine box based on image model matching and recognition described in the present invention, wherein the key feature value includes one or more of a pattern, a brand logo, a color, a size, a shape, a barcode or a QR code.

[0018] On the other hand, the present invention also provides a medicine box storage and access device, which includes:

[0019] The acquisition module acquires one or more faces of the medicine box with obvious features through the first camera in the medicine box identification area and learns the key feature values ​​of each face, and then binds the name and model information of the medicine box through the key feature values ​​to establish a feature database;

[0020] The classification module enters the information of the medicine boxes that need to be replenished, classifies and manages the medicine boxes according to multiple key feature values, and then places the medicine boxes of the same brand and type in different trays according to different specifications;

[0021] The comparison and identification module takes a photo of the medicine boxes in the tray again to determine the type of medicine boxes, and then compares and identifies the medicine boxes in the tray again based on the corresponding medicine box information in the database and the key feature values ​​of the photographed medicine box image. If the comparison results are the same, the medicine boxes are put on the shelf. If the comparison results are different, they are returned to the waiting area.

[0022] The analysis and execution module controls the second camera to take pictures of the pallet and the medicine boxes on the pallet, and uses the checkerboard calibration plate for calibration. The coordinates of the feature points of the calibrated image are superimposed and overlapped with the coordinate relationship of the actual pallet to obtain the two-dimensional coordinates of all the medicine boxes in the pallet in the image; after converting the two-dimensional plane coordinates of the medicine boxes on the image into three-dimensional space coordinates, the shortest route is found according to the three-dimensional coordinates of the target medicine box to pick up the medicine.

[0023] The medicine box storage and access device of the present invention further comprises: a multi-level exposure fusion module, which is used to control the camera to cooperate with one or more fill-in lights around it to continuously shoot multiple images with different exposures of the same surface of the medicine box on the tray, and then pre-process the multiple images with different exposures collected, and register and transform the multiple images with different exposures into the same coordinate system through an image registration algorithm; fuse the multiple registered images through a fusion algorithm to obtain a unique fill-in light parameter fed back from the current medicine box surface; and finally compare the fill-in light parameter with the information of the medicine box in a database to determine the target medicine box;

[0024] A generation module is used to perform deformation calculation according to the height parameter, focal length parameter, viewing angle parameter of the second camera and the laser coefficient of the camera itself, obtain the internal parameter and distortion parameter of the second camera, and generate an image deformation law;

[0025] The correction module is used to correct the panoramic image taken by the second camera to obtain the actual positions of all medicine boxes in the image.

[0026] The medicine box storage and access device of the present invention, wherein the medicine box storage and access device further comprises:

[0027] The acquisition module is used to acquire the identification code of the tray and the medicine box information, associate the identification code of the tray with the information of multiple medicine boxes in the tray, and generate a list of the medicine boxes in the current tray.

[0028] The transport module is used to transport the removed medicine boxes into the medicine basket.

[0029] The beneficial effect of the present invention is that the medicine box retrieval method based on image model matching and recognition and the medicine box storage and retrieval device are cleverly designed, and the key feature values ​​on multiple surfaces of the medicine box are extracted and learned, bound to the medicine box and uploaded to the medicine box database, and then the medicine boxes with relatively high similarity, such as medicine boxes of the same brand and type, are placed in different trays according to different specifications, which can reduce recognition errors, and the medicine boxes in the tray are photographed again by the first camera to determine the type of the medicine box, and then the key feature values ​​of the photographed medicine box image are compared and identified according to the corresponding medicine box information in the database to determine the medicine box in the tray again, so as to determine the type of the medicine box on the tray in advance, and then the tray is put on the shelf, and the second camera obtains the two-dimensional coordinate information of the medicine box on the tray and combines it with the height information of the medicine box through spatial calculation to convert it into the three-dimensional coordinate information of the medicine box, and then the three-dimensional coordinates of the target medicine box are used to find the shortest route for picking up the medicine to ensure accurate grasping of the medicine box, which greatly increases the accuracy of grasping the medicine box. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. The drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work:

[0031] Figure 1 This is a flow chart of a method for taking medicine from a medicine box based on image model matching and recognition according to the first embodiment of the present invention;

[0032] Figure 2 It is a schematic diagram of a medicine box storage and access device according to the second embodiment of the present invention. DETAILED DESCRIPTION

[0033] The terms "first", "second", "third" and "fourth" etc. in the specification and claims of the present invention and the drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0034] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0035] Moreover, the terms "up, down, front, back, left, right, upper end, lower end, longitudinal" and the like indicating directions are all based on the posture and position of the device or equipment described in this solution during normal use.

[0036] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the following will be described clearly and completely in combination with the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are partial embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0037] Embodiment 1:

[0038] A preferred embodiment of the present invention is a method for taking medicine from a medicine box based on image model matching and recognition, such as Figure 1 As shown, a method for taking medicine from a medicine box based on image model matching and recognition includes the following steps:

[0039] S10: In the medicine box identification area, a first camera is used to collect one or more faces of the medicine box with obvious features and learn key feature values ​​on each face, and then the name and model information of the medicine box are bound by the key feature values ​​to establish a feature database;

[0040] The key feature value includes one or more of a pattern, a brand logo, a color, a size, a shape, a barcode or a QR code.

[0041] S20: After entering the information of the medicine boxes that need to be replenished, the medicine boxes are classified and managed according to multiple key feature values ​​of the medicine boxes, and then the medicine boxes of the same brand and type are placed in different trays according to different specifications; for example, the medicine boxes of the same brand and type are placed in different trays according to different specifications, which can reduce recognition errors;

[0042] S30: Take photos of the medicine boxes in the tray again to determine the type of the medicine boxes, and then compare and identify the medicine boxes in the tray again according to the corresponding medicine box information in the database and the key feature values ​​of the photographed medicine box image. If the comparison results are the same, the medicine boxes are put on the shelf. If the comparison results are different, they are returned to the waiting area;

[0043] S40: After the tray is put on the shelf, the second camera is controlled to take pictures of the tray and the medicine boxes on the tray, and the checkerboard calibration plate is used for calibration, and the coordinates of the feature points of the calibrated image are superimposed and overlapped with the coordinate relationship of the actual tray to obtain the two-dimensional coordinates of all the medicine boxes in the tray in the image;

[0044] Furthermore, since the medicine box may have different reflections or overexposure due to its material when photographing it, the medicine box may not be recognized. In order to solve this problem and further improve the accuracy of medicine box recognition, the present invention can combine coordinate recognition and multi-level exposure fusion recognition technology to avoid misrecognition between different medicine boxes in the same tray. The specific contents are as follows: the second camera cooperates with one or more fill lights around it to continuously shoot multiple images with different exposures of the same surface of the medicine box on the tray in a certain sequence; multiple fill lights are arranged around the second camera, and the exposure of the fill lights can be adjusted according to the ambient light each time the medicine box is photographed, thereby ensuring that the exposure of the medicine box is different each time it is photographed.

[0045] Preprocess the collected multiple images with different exposures, and transform the multiple images with different exposures into the same coordinate system through image registration algorithm;

[0046] The registered multiple images are fused through a fusion algorithm to obtain image data of the current surface feedback; the image data is then compared with the information of the medicine box in the database to determine the target medicine box;

[0047] After converting the two-dimensional coordinates of the medicine box on the image into three-dimensional coordinates, the shortest route is found according to the three-dimensional coordinates of the target medicine box to pick up the medicine. By placing similar medicine boxes in different trays according to specifications, it can greatly avoid the mistaken grabbing of similar products on the same tray. Then, through image recognition comparison, coordinate positioning, and multi-level exposure fusion recognition, the accuracy of picking up medicine boxes can be further greatly increased.

[0048] The steps of converting the two-dimensional plane coordinates of the medicine box on the image into three-dimensional space coordinates are as follows:

[0049] The height of the medicine box is identified by the ranging sensor, and then the height of the medicine box and the two-dimensional plane coordinates of the medicine box on the image are spatially calculated to obtain the three-dimensional coordinates of the current medicine box. The method of converting the two-dimensional plane coordinates into three-dimensional space coordinates and the method of forming the path are both very mature technical means in the prior art, and will not be repeated here.

[0050] In one embodiment, the ranging sensor can be set in the identification area, and the height of the medicine box and related key feature values ​​are bound and stored in the feature database in advance. After obtaining the two-dimensional coordinates, the height of the medicine box in the database and the two-dimensional coordinates are automatically calculated spatially to obtain the three-dimensional coordinates; optionally, the ranging sensor can also be set on a pallet or a transport vehicle. After obtaining the two-dimensional coordinates of the medicine box, the height of the medicine box is calculated by the ranging sensor to obtain the three-dimensional coordinates of the medicine box.

[0051] Since the photos taken after being taken are easily deformed, the camera also includes the following steps of camera calibration before taking panoramic photos of the medicine boxes on the tray: deformation calculation is performed according to the second camera height parameter, focal length parameter, viewing angle parameter and the laser coefficient of the second camera itself, the internal parameters and distortion parameters of the second camera are obtained, and the image deformation law is generated; in this embodiment, a series of original image data are first collected, including the second camera height, focal length parameter, viewing angle parameter and the laser coefficient of the second camera itself, and then the second camera height, focal length parameter, viewing angle parameter and the laser coefficient of the second camera itself are adjusted to obtain new image data, and the original image data is compared with multiple groups of image data and the image deformation law is obtained after calculation and processing, and then the camera calibration can be adjusted according to the actual application; it is worth noting that the calculation algorithm involved in this process is a prior art and will not be repeated here. After the second camera takes a panoramic photo of the medicine boxes in the tray, the panoramic image taken by the camera is corrected according to the image deformation law to obtain the actual position of all medicine boxes in the image. The accuracy of grabbing medicine boxes is further improved.

[0052] Furthermore, the tray is provided with a unique identification code; before the medicine box is placed in the tray, the identification code of the tray is obtained and the medicine box information is associated with the identification code to generate a list of medicine boxes in the current tray. By obtaining all the types of medicine boxes on the tray in advance, the range of the camera to identify the medicine box can be greatly narrowed. The camera can identify the type of medicine box within a small known range in the tray and obtain the location information of the medicine box, which has high recognition efficiency and accuracy.

[0053] The medicine box retrieval method based on image model matching and recognition and the medicine box storage and retrieval device are cleverly designed. The key feature values ​​on multiple surfaces of the medicine box are extracted and learned, bound to the medicine box and uploaded to the medicine box database. Then, medicine boxes with relatively high similarity, such as medicine boxes of the same brand and type, are placed in different trays according to different specifications, which can reduce recognition errors. The medicine boxes in the tray are photographed again by the first camera to determine the type of the medicine box. Then, the key feature values ​​of the photographed medicine box image are compared and identified based on the corresponding medicine box information in the database to determine the medicine boxes in the tray again, so as to determine the type of the medicine boxes on the tray in advance. Then, the tray is put on the shelf, and the two-dimensional coordinate information of the medicine boxes on the tray is obtained by the second camera. It is then converted into the three-dimensional coordinate information of the medicine boxes through spatial calculation in combination with the height information of the medicine boxes. Then, the three-dimensional coordinates of the target medicine boxes are used to find the shortest route for picking up the medicine to ensure accurate grabbing of the medicine boxes, which greatly increases the accuracy of grabbing the medicine boxes.

[0054] Embodiment 2:

[0055] The present invention also protects a medicine box storage and access device, such as Figure 2 As shown, including:

[0056] The acquisition module acquires one or more faces of the medicine box with obvious features through the first camera in the medicine box identification area and learns the key feature values ​​of each face, and then binds the name and model information of the medicine box through the key feature values ​​to establish a feature database;

[0057] The classification module enters the information of the medicine boxes that need to be replenished, classifies and manages the medicine boxes according to multiple key feature values, and then places the medicine boxes of the same brand and type in different trays according to different specifications;

[0058] The comparison and recognition module takes a photo of the medicine boxes in the tray again to determine the type of the medicine boxes, and then compares and identifies the medicine boxes in the tray again based on the corresponding medicine box information in the database and the key feature values ​​of the photographed medicine box image. If the comparison results are the same, the medicine boxes are put on the shelf. If the comparison results are different, they are returned to the waiting area. By obtaining all the types of medicine boxes on the tray in advance, the range of the camera to identify the medicine boxes can be greatly narrowed. The type of medicine box can be identified within a known small range in the tray and the location information of the medicine box can be obtained. The recognition efficiency is high and the accuracy rate is high.

[0059] The analysis and execution module controls the second camera to take pictures of the pallet and the medicine boxes on the pallet, and uses the checkerboard calibration plate for calibration. The coordinates of the feature points of the calibrated image are superimposed and overlapped with the coordinate relationship of the actual pallet to obtain the two-dimensional coordinates of all the medicine boxes in the pallet in the image; after converting the two-dimensional plane coordinates of the medicine boxes on the image into three-dimensional space coordinates, the shortest route is found according to the three-dimensional coordinates of the target medicine box to pick up the medicine.

[0060] The medicine box storage and access device also includes:

[0061] The multi-level exposure fusion module is used to control the camera to cooperate with one or more fill-in lights around it to continuously shoot multiple images with different exposures on the same surface of the medicine box on the tray, and then pre-process the multiple images with different exposures collected, and register and transform the multiple images with different exposures into the same coordinate system through an image registration algorithm; fuse the multiple registered images through a fusion algorithm to obtain a unique fill-in light parameter fed back from the current medicine box surface; finally, compare the fill-in light parameter with the information of the medicine box in the database to determine the target medicine box;

[0062] A generation module is used to perform deformation calculation according to the height parameter, focal length parameter, viewing angle parameter of the second camera and the laser coefficient of the camera itself, obtain the internal parameter and distortion parameter of the second camera, and generate an image deformation law;

[0063] The correction module is used to correct the panoramic image taken by the second camera to obtain the actual positions of all medicine boxes in the image.

[0064] The medicine box storage and access device also includes:

[0065] The acquisition module is used to acquire the identification code of the tray and the medicine box information, associate the identification code of the tray with the information of multiple medicine boxes in the tray, and generate a list of the medicine boxes in the current tray.

[0066] The transport module is used to transport the removed medicine boxes to the medicine basket, with a high degree of automation.

[0067] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. A method for taking medicine from a medicine box based on image model matching and recognition, characterized in that: The following steps are involved: In the medicine box identification area, a first camera is used to collect one or more faces of the medicine box with obvious features and learn key feature values ​​on each face, and then the name and model information of the medicine box are bound by the key feature values ​​to establish a feature database; After entering the information of the medicine boxes that need to be replenished, the medicine boxes are classified and managed according to multiple key feature values ​​of the medicine boxes, and then the medicine boxes of the same brand and type are placed in different trays according to different specifications; Take a photo of the medicine boxes in the tray again to determine the type of medicine boxes, and then compare and identify the corresponding medicine box information in the database with the key feature values ​​of the photographed medicine box image to determine the medicine boxes in the tray again. If the comparison results are the same, the medicine boxes are put on the shelf. If the comparison results are different, they are returned to the waiting area. After the pallet is put on the shelf, the second camera is controlled to take pictures of the pallet and the medicine boxes on the pallet. The checkerboard calibration plate is used for calibration, and the coordinates of the feature points of the calibrated image are superimposed and overlapped with the coordinate relationship of the actual pallet to obtain the two-dimensional coordinates of all the medicine boxes in the pallet in the image; after the two-dimensional plane coordinates of the medicine boxes in the image are converted into three-dimensional space coordinates, the shortest route is found according to the three-dimensional coordinates of the target medicine box to pick up the medicine.

2. The method for taking medicine from a medicine box according to claim 1, characterized in that: The following steps are also included before picking up the medicine: The second camera cooperates with one or more fill lights around it to continuously shoot a plurality of images with different exposures on the same surface of the medicine box on the tray; Preprocess the collected multiple images with different exposures, and transform the multiple images with different exposures into the same coordinate system through image registration algorithm; The registered multiple images are fused through a fusion algorithm to obtain the image data of the current surface feedback; the image data is then compared with the information of the medicine box in the database to determine the target medicine box.

3. The method for taking medicine from a medicine box according to claim 1 or 2, characterized in that: The steps for converting the two-dimensional plane coordinates of the medicine box on the image into three-dimensional space coordinates are as follows: The height of the medicine box is identified by a distance measuring sensor, and then the height of the medicine box and the two-dimensional plane coordinates of the medicine box on the image are spatially calculated to obtain the three-dimensional coordinates of the current medicine box.

4. The method for taking medicine from a medicine box according to claim 3, characterized in that: The distance measuring sensor can be arranged in the identification area, and the height of the medicine box and the relevant key feature values ​​can be bound and stored in the feature database in advance.

5. The method for taking medicine from a medicine box according to claim 3, characterized in that: Deformation calculation is performed based on the height parameters, focal length parameters, viewing angle parameters of the second camera and the laser coefficient of the second camera itself to obtain the internal parameters and distortion parameters of the second camera and generate the image deformation law. After the second camera takes a panoramic photo of the medicine boxes in the tray, the panoramic image taken by the camera is corrected according to the image deformation law to obtain the actual positions of all the medicine boxes in the image.

6. The method for taking medicine from a medicine box according to claim 1, characterized in that: The tray is provided with a unique identification code; before the medicine boxes are placed in the tray, the identification code of the tray is obtained and the medicine box information is associated with the identification code to generate a list of the medicine boxes in the current tray.

7. The method for taking medicine from a medicine box according to claim 1, characterized in that: The key feature value includes one or more of a pattern, a brand logo, a color, a size, a shape, a barcode or a QR code.

8. A medicine box storage and retrieval device, characterized in that: include: The acquisition module acquires one or more faces of the medicine box with obvious features through the first camera in the medicine box identification area and learns the key feature values ​​of each face, and then binds the name and model information of the medicine box through the key feature values ​​to establish a feature database; The classification module enters the information of the medicine boxes that need to be replenished, classifies and manages the medicine boxes according to multiple key feature values, and then places the medicine boxes of the same brand and type in different trays according to different specifications; The comparison and identification module takes a photo of the medicine boxes in the tray again to determine the type of medicine boxes, and then compares and identifies the medicine boxes in the tray again based on the corresponding medicine box information in the database and the key feature values ​​of the photographed medicine box image. If the comparison results are the same, the medicine boxes are put on the shelf. If the comparison results are different, they are returned to the waiting area. The analysis and execution module controls the second camera to take pictures of the pallet and the medicine boxes on the pallet, and uses the checkerboard calibration plate for calibration. The coordinates of the feature points of the calibrated image are superimposed and overlapped with the coordinate relationship of the actual pallet to obtain the two-dimensional coordinates of all the medicine boxes in the pallet in the image; after converting the two-dimensional plane coordinates of the medicine boxes on the image into three-dimensional space coordinates, the shortest route is found according to the three-dimensional coordinates of the target medicine box to pick up the medicine.

9. The medicine box storage and access device according to claim 8, characterized in that: The medicine box storage and access device also includes: The multi-level exposure fusion module is used to control the camera to cooperate with one or more fill-in lights around it to continuously shoot multiple images with different exposures on the same surface of the medicine box on the tray, and then pre-process the multiple images with different exposures collected, and register and transform the multiple images with different exposures into the same coordinate system through an image registration algorithm; fuse the multiple registered images through a fusion algorithm to obtain a unique fill-in light parameter fed back from the current medicine box surface; finally, compare the fill-in light parameter with the information of the medicine box in the database to determine the target medicine box; A generation module is used to perform deformation calculation according to the height parameter, focal length parameter, viewing angle parameter of the second camera and the laser coefficient of the camera itself, obtain the internal parameter and distortion parameter of the second camera, and generate an image deformation law; The correction module is used to correct the panoramic image taken by the second camera to obtain the actual positions of all medicine boxes in the image.

10. The medicine box storage and access device according to claim 9, characterized in that: The medicine box storage and access device also includes: An acquisition module, used to acquire the identification code of the tray and the information of the medicine boxes, associate the identification code of the tray with the information of the multiple medicine boxes in the tray, and generate a list of the medicine boxes in the current tray; The transport module is used to transport the removed medicine boxes into the medicine basket.

Images