Machine vision-based filter stick library stocktaking system and stocktaking method

Abstract

The invention discloses a machine vision-based filter stick library stocktaking system and stocktaking method. The stocktaking system comprises an image acquisition system for acquiring goods allocation images of a filter stick library in real time; the image transmission system is used for safely and reliably transmitting the image information acquired by the image acquisition system back to the software system of the server side in real time; the hardware server is used for operating the image acquisition system and the software system; the auxiliary system is used for assisting the software system to quickly realize the state recognition of the cargo image; the software system comprises two subsystems, namely an image processing system and a stacker control system; the image processing system is used for carrying out a series of machine data algorithm processes such as filter stick image RIO region positioning, image segmentation, image preprocessing and image recognition on the goods location image, so that the category and state of the filter stick disc are detected in real time, and a recognition report is generated; and the stacker control system is used for planning the walking route of the stacker.

Description

Technical Field

[0001] The present invention relates to the technical field of filter rod library and disk based on machine vision, and more particularly to a filter rod library and disk system and a library and disk method based on machine vision. Background Art

[0002] Inventory is the most basic daily operation in warehouse management, and its purpose is to find out the actual inventory quantity and make timely data corrections. At present, the filter rod warehouse adopts the manual inventory method, which requires operators to enter the filter rod warehouse area after the power is cut off from the equipment outside the warehouse, and visually count one by one to identify the brand type of filter rods in each cargo position, the status of each brand box group (empty tray / full tray), the number of empty cargo positions, manually record the data, and compare it with the data in the management system to complete the inventory operation. Operators reported that it is difficult for the human eye to judge the brand type and box group status of filter rod trays on high-rise shelves. In order to avoid omissions, repeated statistics, brand judgment errors and other mistakes during the inventory process, multiple re-inspections are required, and the overall inventory time is relatively long.

[0003] Currently, mobile robot and drone inventory systems already exist. However, current robot inventory requires structural modification of the warehouse, which is difficult and costly. It can only conduct an inventory of the entire warehouse area, has a single function, and cannot be compared with the data of the existing filter rod warehouse management system. It cannot meet the diversified inventory management needs of the filter rod elevated warehouse on site. Summary of the invention

[0004] The present invention aims to solve the deficiencies in the prior art. In view of the high labor cost, great management difficulty, inconvenient inventory, and other problems existing in the inventory of filter rod warehouses, a filter rod warehouse inventory system and inventory method based on machine vision are provided. The system realizes automatic inventory of filter rod warehouses based on machine vision technology, and there is no need for large-scale transformation of the filter rod warehouse automation warehouse. This is of great significance for improving inventory efficiency, reducing labor intensity, and improving the safety of warehouse operation.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: A filter rod storage system based on machine vision, comprising: Image acquisition system: used to obtain the image of the filter rod storage in real time; Image transmission system: a software system used to transmit the image information obtained by the image acquisition system back to the server in a safe, reliable and real-time manner; Hardware server: used to run the image acquisition system and software system; Auxiliary system: used to assist the software system to quickly realize the status recognition of cargo images; The software system includes two subsystems: image processing system and stacker control system; Image processing system: The image processing system is used to perform a series of machine data algorithm processes such as filter rod image RIO area positioning, image segmentation, image preprocessing, and image recognition on the cargo image, so as to realize real-time detection of filter rod tray category and status, and generate recognition reports; Stacker control system: used for stacker's travel route planning and automatic control of the stacker's motion system.

[0006] In some embodiments, the image resolution acquired by the image acquisition system should be greater than or equal to 1920X1080P.

[0007] In some embodiments, the image transmission system should have a delay of less than 50ms when transmitting data.

[0008] In some embodiments, the auxiliary system includes a lighting system. When the ambient light in the filter rod library is lower than 1000 lux, the lighting system is used to supplement the ambient illumination to 3000 lux-5000 lux.

[0009] In some embodiments, the auxiliary system also includes a pallet auxiliary positioning pattern, the pallet auxiliary positioning pattern includes a digital pattern and a line pattern, the digital pattern corresponds to the filter rod category, and the line pattern is used for image segmentation during image recognition.

[0010] This embodiment also provides a filter rod library method based on machine vision, which is implemented based on any of the above systems and includes the following steps: S101 Establish the working conditions of the inventory system. Before the system works, the basic inventory system working environment should be established. The specific steps include: The first step is to selectively fill in the light according to the ambient illumination of the filter rod warehouse. First, use an illuminance meter to measure the illumination value of the filter rod warehouse. When the illumination is lower than 1000lux, use COB-type LED floodlights for fill-in lighting to ensure the consistency of the filter rod warehouse cargo position and ambient illumination. The number of floodlights can be installed and laid out according to the actual situation on site. When the ambient illumination reaches 3000lux-5000lux, it means that the illumination is appropriate and no more needs to be added. The second step is to paste auxiliary patterns on the filter rod disc. To ensure the reliability of image processing, the four corners of the filter rod disc should be labeled; S102 system basic parameter settings, including: The first step is to set the basic information of the filter rod warehouse, including the number of filter rod warehouse cargo spaces, the number of layers, and the number of stackers; The second step is to set the image processing parameters, including the basic parameters of the camera's resolution, frame rate, and the number of image storages; The third step is to set the cargo location to be tested, directly select the row and column number through the system interface, or check one or several types of filter rods; S103 filter rod library disk library mode setting, the specific steps include: The first step is to select the stacker operation mode according to the number of stackers set in S101, including single stacker horizontal, single stacker vertical, double stacker horizontal, and double stacker vertical. The operation time of single and double stackers is different, and users can select according to the actual inventory time requirements. When the double stacker has an interlocking anti-collision device, consider using it. In the second step, according to the cargo position to be detected in the basic parameter setting of the system in S102, an inventory method is selected; the system inventory method of the present invention is as follows Figure 5 shown.

[0011] S104 generates a route for inventory counting. After the settings of S101, S102 and S103, the inventory counting system will automatically generate an inventory counting route according to the inventory counting method and the information of the cargo locations to be inspected. Meanwhile, the inventory counting system can also modify the information of the cargo locations to be inspected during operation. The system will re-plan the inventory counting route and display the inventory counting route map. S105 filter rod warehouse cargo location image acquisition, after the system is set up and the inventory route is generated, the inventory system starts working. The image acquisition steps include: In the first step, according to the coordinate information of the cargo position to be detected, the stacker automatically moves to the target cargo position through the control command of the stacker control system; In the second step, the camera takes pictures of the filter rod storage location, collects 30 images at a time, and stays for 2 seconds to ensure the reliability of image acquisition; S106 image transmission: after the image acquisition system acquires the image, the image is transmitted through the on-site optical fiber + network cable wired image transmission system to acquire the real-time image to the hardware server software end; S107 filter rod image recognition: after the image processing system acquires the image, it uses a variety of image processing algorithms to recognize the filter rod disk image; The first step is image preprocessing. After acquiring the image, in order to ensure the stability of subsequent image processing, 30 frames of images need to be preprocessed. The present invention first uses a Gaussian filter algorithm to remove image noise, and then uses the MSRCR algorithm to process the image, fully eliminating the influence of ambient light on the filter rod disk, and improving the stability of the image recognition method; The second step is image segmentation. After preprocessing the image, the image is segmented. The line pattern and the digital pattern are segmented using the Canny edge detection method, and the RIO area of ​​the digital pattern and the RIO area of ​​the line are located respectively. The third step is to identify the state of the filter rod and disk. After segmenting the filter rod and disk RIO area, this part of the image is identified and processed; S108 outputs the current inventory result. After the inventory system recognizes the status of the current cargo location after step S107, the system automatically stores the original image taken by the camera, the segmented image during the image processing, the recognized image, and the inventory result; S109 updates the inventory route. After outputting the current inventory result, the system records the locations that have been counted and the locations that have not been counted, automatically updates the inventory route, and controls the stacker to go to the next target location for inventory; S110 outputs inventory report. After the stacker has traversed all the cargo locations to be tested, the filter rod inventory system based on machine vision will automatically generate an inventory report. The report includes a variety of inventory information such as the number of items counted, the type of goods, the status of the goods, and the time of inventory.

[0012] The filter rod library inventory system and inventory method provided in this application based on machine vision have beneficial effects including but not limited to: Compared with the background technology, the present invention has the following beneficial effects: compared with the inventory by drones and mobile robots, the present invention does not require large-scale transformation of the existing warehouse of the filter rod warehouse, and uses the existing stacker in the filter rod warehouse as a carrier for development, which can greatly reduce the development cost and R&D cycle. At the same time, the present invention adopts a variety of machine vision technologies, and the system has good stability and reliability, which can effectively replace the existing manual visual inspection method and realize the rapid inventory of the cargo positions in the filter rod warehouse. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 System structure and data transmission relationship diagram; Figure 2 Auxiliary positioning pattern diagram of auxiliary system; Figure 3 Image recognition method process; Figure 4 Workflow diagram of filter rod library and tray library system based on machine vision; Figure 5 Stacker operation diagram Figure 6 Schematic diagram of stacker operation mode DETAILED DESCRIPTION

[0014] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described in more detail below in conjunction with the drawings in the preferred embodiments of the present application. In the drawings, the same or similar reference numerals throughout represent the same or similar parts or parts with the same or similar functions. The described embodiments are part of the embodiments of the present application, not all of the embodiments. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limitations on the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0015] The following will be combined Figure 1-6 It is worth noting that the following embodiments are only used to explain the present application and do not constitute a limitation on the present application.

[0016] Example:

[0017] The inventory system mainly includes two aspects: hardware system and software system.

[0018] The hardware system includes: image acquisition system, image transmission system, hardware server and auxiliary system.

[0019] Image acquisition system: installed on the stacker rack or frame, used to obtain the cargo image of the filter rod warehouse in real time. Because the present invention uses an image algorithm based on machine vision to process the real-time image, in order to ensure the reliability and stability of subsequent image processing, the image resolution obtained by the image acquisition system is greater than or equal to 1920X1080P.

[0020] Image transmission system: A software system used to transmit the image information obtained by the image acquisition system back to the server in a safe, reliable and real-time manner. Wired image transmission or wireless image transmission solutions can be used. A further solution of the present invention is to use an optical communication + network cable transmission + switch solution to ensure the long-term reliability of image transmission to meet the low latency requirements of the image. The delay should be less than 50ms.

[0021] Hardware server: used to run image acquisition programs, image processing systems, stacker control systems, network communications and other functions. It is the physical carrier for the software system to perform comprehensive information processing. A further solution of the present invention is to use a high-performance PC server computer as a hardware server to meet the various information processing needs of the inventory system.

[0022] Auxiliary system: used to assist the image processing system to quickly realize the status recognition of the cargo image, which mainly includes the lighting system and the auxiliary positioning pattern of the pallet. A further solution of the present invention is that when the ambient light in the filter rod warehouse is lower than 1000 lux, the filter rod warehouse environment should be supplemented with light. To ensure the reliability and stability of image processing, a COB-type LED supplementary light can be used to supplement the ambient illumination to 3000lux-5000lux to ensure the effectiveness of image acquisition. In order to further improve the real-time and stability of image segmentation during image processing, the following is invented: Figure 2 The auxiliary positioning pattern shown includes a digital pattern and a line pattern, the digital pattern corresponds to the filter rod category, and the line pattern is used for image segmentation during image recognition. The pattern of the present invention is posted on the four corners of the filter rod tray to ensure that the image processing system can quickly recognize the filter rod library status and the cargo type.

[0023] The software system includes two subsystems: image processing system and stacker control system.

[0024] Image processing system: includes multiple processes such as image acquisition, image storage, and image recognition. After the image processing system transmits the image of the filter rod in the cargo area obtained by the image acquisition system to the software system via the image transmission system, the image processing system performs a series of machine data algorithm processes such as filter rod image RIO area positioning, image segmentation, image preprocessing, and image recognition on the image, so as to realize real-time detection of the type and status of the filter rod tray and generate an identification report.

[0025] Stacker control system: used for stacker travel route planning and automatic control of the stacker motion system. The stacker motion system is the stacker's own on-site control system, which is responsible for the basic motion control of the stacker, including start, stop, and movement to a specified position. During the operation of the filter rod warehouse inventory system, the stacker control system sends control commands to the stacker motion system to drive the stacker to traverse the cargo positions to be inspected upward, downward, left, and right in sequence. During this process, the stacker leads the camera to "walk" in the target cargo positions in turn to perform image recognition and inventory the cargo positions to be inspected.

[0026] The present invention proposes three inventory methods, namely, full inventory, designated location inventory, and category inventory. Full inventory: that is, all cargo locations in the filter rod warehouse are inventoried one by one and an inventory report is generated. Designated location inventory: that is, the user or operator sets the cargo location coordinate information that needs to be inventoried in the software system, such as one or more cargo locations in the row and column, and the inventory system will automatically generate an inventory route to realize the cargo location inventory. Category inventory: that is, the user or operator sets one or several filter rod varieties that need to be inventoried in the software system, such as one, two, or three types, and the inventory system will automatically generate an inventory route to realize the cargo location inventory.

[0027] The present invention proposes four stacker operation modes, including single stacker horizontal, single stacker vertical, double stacker horizontal, and double stacker vertical. The present invention mainly considers the distribution of stackers in different types of filter rod warehouses, and invents and designs four stacker operation modes in the software information system. Users or operators can make comprehensive choices based on the actual operation of the filter rod warehouse and the real-time requirements of the filter rod warehouse.

[0028] See also Figure 1 , Figure 1 It is a diagram of system structure and data transmission relationship; The filter rod library system structure of the present invention is as follows: Figure 1 The system is divided into two parts: hardware system and software system. The auxiliary system includes the auxiliary positioning pattern proposed by the present invention such as Figure 2 As shown, the two parts of the image acquisition system and the fill light according to the ambient illumination (the fill light is filled according to the ambient illumination) provide the basic conditions for image acquisition and subsequent image processing. The image acquisition system uses a high-definition camera with a resolution of 1080P or above to acquire images in MP4 format, and transmits the acquired real-time images to the hardware server on the PC through the image transmission system to complete image decoding, image storage, and image format conversion, and then converts the MP4 format image into the standard format of 1920X1080, JPEG to the image processing software system. The software system uses the image processing method of the present invention ( Figure 3 The filter rod type and quantity can be identified and the inventory report of the current location can be output.

[0029] In order to realize the identification of all cargo locations, the present invention also provides a stacker operation control system (software system), which is a part of the software system of the present invention, and is used to realize the control of the stacker system to realize the image acquisition of multiple cargo locations to meet different inventory requirements. The system needs to be linked with the stacker motion control system. The stacker motion system is the motion control system of the stacker itself, which is used to control the stacker's start, stop, cargo location movement and other functions. The stacker control system obtains the position information of the stacker motion system through the PROFINET communication protocol and the RJ-45 interface, updates the feedback cargo location coordinate information in real time, and sends control commands to the stacker motion system through the USART interface to control the stacker to move to the target cargo location and obtain the image of the target cargo location.

[0030] See also Figure 2 , Figure 2 To assist in positioning the pattern; The line pattern in the auxiliary positioning pattern is used to help the image processing system quickly locate the edge of the cargo space, and the digital pattern is used to distinguish different types of filter rods. The type of filter rod is identified by identifying the digital pattern during the image processing process.

[0031] See also Figure 3 , Figure 3 Image recognition method process After obtaining the original JPEG cargo image with a resolution of 1920X1080, the image is first preprocessed. The present invention uses a Gaussian algorithm to denoise the image, and then uses the MSRCR method to eliminate the adverse effects of light on the image, better restore the image details, and realize image preprocessing. After preprocessing, the Canny edge detection image is used to obtain the edge features of the cargo image to achieve the elimination of the non-RIO area and the positioning of the RIO area. After obtaining the RIO area, the desent algorithm model trained in advance is used to perform feature recognition on the image to be tested, and the numbers of the auxiliary positioning pattern and the circular surface of the filter rod image (the filter rod image is a pile of circles stacked together) are respectively identified to achieve statistics on the types and quantities of filter rods, and finally output a recognition report.

[0032] refer to Figure 1 , which is a filter rod storage and pallet storage system based on machine vision disclosed in one embodiment of the present invention, wherein the system uses a stacker to sequentially traverse the target cargo positions and uses a machine vision algorithm for image recognition to realize the recognition of the status of the filter rod cargo positions. The implementation methods of the various parts of the system are as follows: The image acquisition system 101 is used to acquire the image of the filter rod tray in the filter rod warehouse, and uses a 1920X1080P industrial camera to ensure the reliability of image acquisition. In addition, to ensure the reliability of camera installation, a corresponding camera mounting bracket should be designed according to the physical structure of the stacker to ensure that the camera is firm and does not shake when the stacker is "walking" to acquire images.

[0033] The image transmission system 102 is used to transmit images to the hardware server. In actual implementation, the system should fully consider the image transmission bandwidth problem. To ensure low latency and reliability of image transmission, optical communication + network cable transmission is used for limited transmission, and the network cable is laid out with CAT6A and above 10 Gigabit network cables.

[0034] Hardware server 103 is the operating platform for software systems such as graphics processing algorithms and stacker control systems. When selecting a server, a computer server with Intel 11th generation i7 or above performance should be used, with a running memory greater than 64GB and a 1TB PCIE4.0 solid-state hard drive as the basic hardware configuration to ensure the stability of the software system and sufficient computing power.

[0035] After the image is transmitted to the hardware server by the image processing system 104 and the image transmission system 103, the image processing system needs to process the image. The image processing process is as follows: Figure 3As shown in the figure, the specific process includes using Gaussian algorithm + MSRCR method to preprocess the image, combining with Canny edge detection method to locate the RIO area, and finally using Densent algorithm model to recognize the image, so as to realize accurate recognition of the filter rod warehouse storage location image, and identify the number and type of filter rods (digital recognition).

[0036] The stacker control system 105 is physically connected to the stacker motion system through the USART interface, and sends instructions to control the motion system to drive the stacker. During the inventory process, the stacker control system will automatically plan the collection route and control the image collection system in turn to complete the acquisition of real-time images at the cargo locations to be measured.

[0037] Auxiliary system, in order to ensure the reliability of image acquisition, due to full consideration of the ambient illumination problem, a digital illuminance meter can be used to measure the ambient illumination. When the ambient illumination is lower than 1000 lux, a COB type LED fill light should be used for illumination fill light. In order to ensure the real-time and reliability of image segmentation, the four corners of the filter rod plate should also be pasted with Figure 2 The auxiliary positioning pattern shown can be scaled according to the size of the filter rod disc, but the length of the line pattern should be greater than 5CM, and the height of the digital pattern should be greater than 3CM. The digital pattern can be used to define the type of filter rod disc to distinguish different types of filter rods. This digital pattern can be used to define the type of filter rod according to actual production.

[0038] This reference Figure 4 , is a flowchart of a filter rod library and tray library system based on machine vision according to an embodiment of the present invention, and the workflow includes the following steps: S101 Establish the working conditions of the inventory system. Before the system works, the basic inventory system working environment should be established. The specific steps include: The first step is to selectively fill in the light according to the ambient illumination of the filter rod warehouse. First, use an illuminance meter to measure the illumination value of the filter rod warehouse. When the illumination is lower than 1000lux, use COB-type LED floodlights for fill-in lighting to ensure the consistency of the filter rod warehouse cargo position and ambient illumination. The number of floodlights can be installed and laid out according to the actual situation on site. When the ambient illumination reaches 3000lux~5000lux, it means that the illumination is appropriate and no more needs to be added.

[0039] The second step is to paste auxiliary patterns on the filter rod disc. To ensure the reliability of image processing, the four corners of the filter rod disc should be labeled and pasted. In the actual construction process, manual pasting can be used when the empty filter rod disc is recovered.

[0040] S102 Basic parameter setting of the system. When designing the present invention, the actual differences between different filter rod libraries are fully considered. Before the filter rod library system is operated, the basic parameters of the system need to be set.

[0041] The first step is to set the basic information of the filter rod warehouse, including the number of filter rod warehouse storage locations, the number of layers, and the number of stackers (one or two).

[0042] The second step is to set the image processing parameters, including basic parameters such as camera resolution, frame rate, and number of images to be stored.

[0043] The third step is to set the cargo location to be tested. You can directly select the row and column number through the system interface, or check one or several types of filter rods.

[0044] S103 filter rod warehouse storage method setting, this step needs to be set comprehensively according to the number of stackers and storage methods, and its specific steps include.

[0045] The first step is to select the stacker operation mode according to the number of stackers set in S101. The specific stacker operation mode is as follows: Figure 6 As shown, it includes single stacker horizontal, single stacker vertical, double stacker horizontal, and double stacker vertical. The operation time of single and double stackers is different. Users can choose according to the actual inventory time requirements. However, during the implementation process, it should be noted that there may be a risk of collision when collecting data horizontally with double stackers. When the double stacker has an interlocking anti-collision device, it can be considered to use it. At the same time, the system software level has been designed to fully avoid the risk of collision.

[0046] In the second step, according to the cargo position to be detected in the basic parameter setting of the system in S102, an inventory method is selected. The system inventory method of the present invention is as follows: Figure 4 shown.

[0047] S104 generates the inventory route. After the settings of S101, S102, and S103, the inventory system will automatically generate the inventory route according to the inventory method and the information of the cargo locations to be inspected. At the same time, the inventory system can also modify the information of the cargo locations to be inspected during operation. The system will re-plan the inventory path and display the inventory route map.

[0048] S105: Image acquisition of filter rod warehouse cargo positions. After the system is set up and the inventory route is generated, the inventory system starts working. The image acquisition steps include:

[0049] In the first step, according to the coordinate information of the cargo location to be detected, the stacker automatically moves to the target cargo location through the control command of the stacker control system.

[0050] In the second step, the camera takes pictures of the filter rod warehouse storage location images, collects 30 images at a time, and stays for 2 seconds to ensure the reliability of image acquisition.

[0051] S106 Image transmission: After the image acquisition system acquires the image, the image is transmitted through the on-site optical fiber + network cable wired image transmission system to acquire the real-time image to the hardware server software end.

[0052] S107 filter rod image recognition: after the image processing system acquires the image, it uses a variety of image processing algorithms to recognize the filter rod disk image.

[0053] The first step is image preprocessing. After acquiring the image, in order to ensure the stability of subsequent image processing, 30 frames of images need to be preprocessed. The present invention first uses a Gaussian filter algorithm to remove image noise, and then uses the MSRCR algorithm to process the image, fully eliminating the influence of ambient light on the filter rod disk and improving the stability of the image recognition method.

[0054] The second step is image segmentation. After preprocessing the image, the image is segmented. Since the present invention adopts an auxiliary positioning pattern, the line pattern and the digital pattern can be segmented using the Canny edge detection method to locate the RIO area of ​​the digital pattern and the RIO area of ​​the line respectively. The third step is to identify the state of the filter rod tray. After segmenting the filter rod tray RIO area, the partial image is identified and processed. The present invention adopts the Densent algorithm model for identification, which can quickly identify the number of filter rods in the filter rod tray and the type of filter rods within 2 seconds.

[0055] S108 outputs the current inventory result. After the inventory system recognizes the status of the current cargo location after step S107, the system will automatically store the original image taken by the camera, the segmented image during the image processing process, the recognized image, and the inventory result.

[0056] S109 updates the inventory route. After outputting the current inventory result, the system records the locations that have been counted and the locations that have not been counted, automatically updates the inventory route, and controls the stacker to go to the next target location for inventory.

[0057] S110 outputs inventory report. After the stacker has traversed all the cargo locations to be tested, the filter rod inventory system based on machine vision will automatically generate an inventory report. The report includes inventory information such as the number of items counted, the type of goods, the status of the goods, and the time of inventory.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A filter rod storage system based on machine vision, characterized in that: include Image acquisition system: used to obtain the image of the filter rod storage in real time; Image transmission system: a software system used to transmit the image information obtained by the image acquisition system back to the server in a safe, reliable and real-time manner; Hardware server: used to run the image acquisition system and software system; Auxiliary system: used to assist the software system to quickly realize the status recognition of cargo images; The software system includes two subsystems: image processing system and stacker control system; Image processing system: The image processing system is used to perform a series of machine data algorithm processes such as filter rod image RIO area positioning, image segmentation, image preprocessing, and image recognition on the cargo image, so as to realize real-time detection of filter rod tray category and status, and generate recognition reports; Stacker control system: used for stacker's travel route planning and automatic control of the stacker's motion system.

2. A filter rod storage system based on machine vision according to claim 1, characterized in that: The image resolution acquired by the image acquisition system should be greater than or equal to 1920X1080P.

3. A filter rod storage system based on machine vision according to claim 1, characterized in that: When the image transmission system transmits data, the delay should be less than 50ms.

4. The filter rod storage system based on machine vision according to claim 1, characterized in that: The auxiliary system includes a lighting system. When the ambient light in the filter rod library is lower than 1000 lux, the lighting system is used to supplement the ambient illumination to 3000 lux-5000 lux.

5. A filter rod storage system based on machine vision according to claim 4, characterized in that: The auxiliary system also includes a pallet auxiliary positioning pattern, which includes a digital pattern and a line pattern. The digital pattern corresponds to the filter rod category, and the line pattern is used for image segmentation during image recognition.

6. A filter rod library method based on machine vision, characterized in that: The system implementation according to any one of claims 1 to 5 comprises the following steps: S101 Establish the working conditions of the inventory system. Before the system works, the basic inventory system working environment should be established. The specific steps include: The first step is to selectively fill in the light according to the ambient illumination of the filter rod warehouse. First, use an illuminance meter to measure the illumination value of the filter rod warehouse. When the illumination is lower than 1000lux, use COB-type LED floodlights for fill-in lighting to ensure the consistency of the filter rod warehouse cargo position and ambient illumination. The number of floodlights can be installed and laid out according to the actual situation on site. When the ambient illumination reaches 3000lux-5000lux, it means that the illumination is appropriate and no more needs to be added. The second step is to paste auxiliary patterns on the filter rod disc. To ensure the reliability of image processing, the four corners of the filter rod disc should be labeled; S102 system basic parameter settings, including: The first step is to set the basic information of the filter rod warehouse, including the number of filter rod warehouse cargo spaces, the number of layers, and the number of stackers; The second step is to set the image processing parameters, including the basic parameters of the camera's resolution, frame rate, and the number of image storages; The third step is to set the cargo location to be tested, directly select the row and column number through the system interface, or check one or several types of filter rods; S103 filter rod library disk library mode setting, the specific steps include: The first step is to select the stacker operation mode according to the number of stackers set in S101, including single stacker horizontal, single stacker vertical, double stacker horizontal, and double stacker vertical. The operation time of single and double stackers is different, and users can select according to the actual inventory time requirements. When the double stacker has an interlocking anti-collision device, consider using it. Step 2: Select the inventory method according to the cargo location to be inspected in the basic parameter settings of the S102 system; S104 generates a route for inventory counting. After the settings of S101, S102 and S103, the inventory counting system will automatically generate an inventory counting route according to the inventory counting method and the information of the cargo locations to be inspected. Meanwhile, the inventory counting system can also modify the information of the cargo locations to be inspected during operation. The system will re-plan the inventory counting route and display the inventory counting route map. S105 filter rod warehouse cargo location image acquisition, after the system is set up and the inventory route is generated, the inventory system starts working. The image acquisition steps include: In the first step, according to the coordinate information of the cargo position to be detected, the stacker automatically moves to the target cargo position through the control command of the stacker control system; In the second step, the camera takes pictures of the filter rod storage location, collects 30 images at a time, and stays for 2 seconds to ensure the reliability of image acquisition; S106 image transmission: after the image acquisition system acquires the image, the image is transmitted through the on-site optical fiber + network cable wired image transmission system to acquire the real-time image to the hardware server software end; S107 filter rod image recognition: after the image processing system acquires the image, it uses a variety of image processing algorithms to recognize the filter rod disk image; The first step is image preprocessing. After acquiring the image, in order to ensure the stability of subsequent image processing, 30 frames of images need to be preprocessed. The present invention first uses a Gaussian filter algorithm to remove image noise, and then uses the MSRCR algorithm to process the image, fully eliminating the influence of ambient light on the filter rod disk, and improving the stability of the image recognition method; The second step is image segmentation. After preprocessing the image, the image is segmented. The line pattern and the digital pattern are segmented using the Canny edge detection method, and the RIO area of ​​the digital pattern and the RIO area of ​​the line are located respectively. The third step is to identify the state of the filter rod and disk. After segmenting the filter rod and disk RIO area, this part of the image is identified and processed; S108 outputs the current inventory result. After the inventory system recognizes the status of the current cargo location after step S107, the system automatically stores the original image taken by the camera, the segmented image during the image processing process, the recognized image, and the inventory result; S109 updates the inventory route. After outputting the current inventory result, the system records the locations that have been counted and the locations that have not been counted, automatically updates the inventory route, and controls the stacker to go to the next target location for inventory; S110 outputs inventory report. After the stacker has traversed all the cargo locations to be measured, the filter rod inventory system based on machine vision will automatically generate an inventory report. The report includes a variety of inventory information such as the number of items counted, the type of goods, the status of the goods, and the time of inventory.

Images