A real-time monitoring and early warning method and system for key impurities based on optoelectronic impurity removal of cut tobacco

Abstract

The present invention relates to the field of tobacco impurity removal, and proposes a real-time monitoring and early warning method and system for key impurities based on optoelectronic impurity removal of cut tobacco. First, scan the tobacco leaves and collect images to obtain high-quality tobacco leaf images. Second, determine whether there are impurities in the tobacco leaf images. If there are, they are used as impurity images. Third, cache continuous multiple segments of impurity images, splice the continuous multiple segments of impurity images into a complete impurity image through timestamps, and upload it to the upper industrial control computer. Fourth, the upper industrial control computer monitors the complete impurity image in real time. The image is displayed on the statistical dashboard, and key impurities are identified based on the AI classification algorithm, and it is judged whether the impurity belongs to a key impurity. If it belongs to a key impurity, an audible and visual alarm is issued. The present invention realizes the real-time monitoring and early warning of key impurities, and solves the technical problem that the existing technology is in a black box state for tobacco impurity removal and cannot perform visual processing on impurity information.

Description

Technical Field

[0001] The present invention relates to the field of tobacco impurity removal, and in particular to a real-time monitoring and early warning method and system for key impurities based on optoelectronic impurity removal of cut tobacco. Background Art

[0002] In the field of tobacco impurity removal, the occurrence of key impurities in tobacco has always been the most concerned issue for cigarette manufacturers. Frequent quality incidents of key impurities in the manufacturing process of cigarette manufacturers have caused huge losses. The appearance of key impurities in tobacco not only means that the purity of tobacco will be greatly affected, but also may mean that major failures may occur in tobacco production equipment (for example, when there is serious misalignment and friction between the production line belt and the connecting parts, the glue particles generated by the belt friction and the paint on the connecting parts will be wiped off, and the glue particles and paint blocks will directly fall into the normal tobacco material and enter the subsequent processes, as well as plastics and rubbers accidentally mixed in the previous processes. These impurities will have a significant impact on the taste of cigarettes. Therefore, glue particles, paint blocks, plastics, rubbers, etc. that produce abnormal odors belong to key impurities, and as long as some of these key impurities appear, they will appear continuously).

[0003] If key impurities are found on the production line, tobacco manufacturers will stop the operation of the entire production line, and then arrange a large number of personnel to conduct centralized inspections in the previous and subsequent sections where the key impurities are found. On the one hand, it is to prevent key impurities from entering the subsequent normal production line and causing a decline in the quality of cigarette products. On the other hand, it is to check whether major failures have occurred in the production equipment, which has led to the appearance of key impurities. This kind of manual inspection is time-consuming and laborious, seriously affecting the production efficiency of tobacco manufacturers, and at the same time causing a large amount of material waste.

[0004] At present, the foreign and domestic impurity removal machines cannot give an alarm when detecting key impurities. The working principle is that after the images collected by the line array / area array camera are transmitted to the image processing unit, they will be cached in the internal storage first (only a small amount of data can be cached each time), and then the impurities in the tobacco leaves will be identified through the image processing function of the image processing unit, and then the impurity signal will be transmitted to the removal mechanism. After the removal mechanism completes the removal action, the FPGA will clear the previously cached data and receive the subsequent collected pictures. For example, a kind of intelligent optoelectronic impurity removal method based on broken tobacco processing proposed in patent CN116965581A. Because the tobacco leaves move at a relatively fast speed during the transmission process, the images collected by the line array / area array camera can only capture part of the impurities and cannot capture the complete impurities. At the same time, the optoelectronic impurity removal system completes the detection, identification and removal of impurities in an extremely short time (<4ms), and the whole process is in a black box state. During the whole process, we only know that the system has identified the impurities and completed the removal action, but we don't know the specific information of the impurities (size, category, whether they are key impurities), and it is impossible to perform visual processing on the impurity information, including the real-time images, sizes, categories and quantities of the impurities, let alone give a real-time warning for key impurities and conduct subsequent mining of impurity data. Summary of the Invention

[0005] In order to solve the technical problems that the existing technology cannot visualize the information of key impurities and realize the real-time warning of impurities, the present invention proposes a real-time monitoring and warning method and system for key impurities based on optoelectronic impurity removal of cut tobacco, so as to realize the real-time monitoring and alarm of key impurities and give a warning of possible equipment failures.

[0006] The specific solutions are as follows:

[0007] A real-time monitoring and warning method for key impurities based on optoelectronic impurity removal of cut tobacco includes the following steps:

[0008] Step 1: Scan the tobacco leaves and collect images: In the optoelectronic impurity removal system of cut tobacco, use a line array or area array camera to scan the tobacco leaves and collect the image information of the tobacco leaves;

[0009] Step 2: Identify impurities: Judge whether there are impurities in the tobacco leaf image. If so, use it as an impurity image;

[0010] Step 3: Stitch the impurity images: Cache continuous multiple segments of impurity images and record the timestamps until no impurities appear in the next image. Then stitch the continuous multiple segments of impurity images into a complete impurity image through the timestamps, and upload the complete impurity image to the upper industrial control computer by using a switch and a gigabit network;

[0011] Step 4: Real-time monitoring and early warning of key sundries: The upper industrial control computer monitors the complete sundry images in real time. The images are displayed on the statistical dashboard. Based on the AI classification algorithm, key sundries are identified, and it is determined whether the sundry belongs to key sundries. If it belongs to key sundries, an audible and visual alarm is issued.

[0012] Preferably, the identification of sundries in Step 2 and the splicing and integration of sundry images in Step 3 are implemented in the image processing unit. The image processing unit identifies the image information and determines whether it is a sundry image, and caches continuous multiple segments of sundry images, which are spliced into a complete sundry image through image splicing technology, and then the sundry image is transmitted to the gigabit switch. The gigabit switch uses the gigabit network to transmit the sundry image to the upper industrial control computer.

[0013] Preferably, the method of splicing sundry images in Step 3 is as follows:

[0014] Step 3.1: Construct two image buffers, one is the defect image buffer and the other is the complete image buffer;

[0015] Step 3.2: After the image detection starts, the first identified sundry image is first stored in the defect image buffer. If the sundry appears in at least one sundry image, all sundry images with the sundry are sequentially stored in the defect image buffer; continuously store the consecutive images of the complete sundry until the last image is not a sundry image, which means a splicing task is completed;

[0016] Step 3.3: According to the timestamps recorded in the sundry images, all sundry images in the defect image buffer are spliced into a complete sundry image and copied to the complete image buffer, and the complete sundry image is uploaded to the upper computer through the complete image buffer; and the defect image buffer is emptied until the next sundry image enters, and then repeat Step 3.2.

[0017] Preferably, the method for identifying key sundries in Step 4 is as follows:

[0018] Use YOLOv5 in AI technology to build a model; the YOLOv5 model is used to identify and classify sundries, and determine whether the sundry belongs to key sundries. If it belongs to key sundries, the warning module controls the audible and visual alarm to issue a warning. If the same type of key sundries continuously appear within the same time period, the warning module issues a key sundry warning prompt, that is, the sound decibel of the audible and visual alarm increases, and the flashing frequency of the red light accelerates;

[0019] At the same time, the category, size, and quantity of the sundry are displayed in real time on the key sundry statistical dashboard, and it is determined whether the sundry is caused by a production equipment failure.

[0020] Preferably, the key debris analysis and warning steps can also: determine whether the key debris appears continuously within a period of time. If the debris appears continuously, the system will issue a key debris warning prompt.

[0021] A key debris warning system based on optoelectronic impurity removal of sliced tobacco

[0022] It includes: a sliced tobacco optoelectronic impurity removal system that identifies impurity images, and a host industrial control computer that classifies, warns, and displays the impurity images;

[0023] The sliced tobacco optoelectronic impurity removal system includes: an optical imaging module that captures tobacco leaf images, an image acquisition module that acquires tobacco leaf images, an image processing unit that identifies impurities and splices impurity images in the tobacco leaf images, and an ejection actuator that ejects impurities based on solenoid valves;

[0024] The host industrial control computer includes: a key debris analysis and processing module that identifies key debris based on an AI classification algorithm, and a key debris statistics dashboard that displays key debris images.

[0025] Preferably, the optical imaging module is composed of at least one line array or area array camera, a lens, and a light source;

[0026] The image acquisition module includes an image acquisition card. The image acquisition card acquires tobacco leaf image information to generate a tobacco leaf image, and further transmits the tobacco leaf image to the image recognition sub-module of the image processing module;

[0027] The image recognition sub-module processes the acquired image information to identify whether there are impurities in the tobacco leaf image. If there are impurities, the tobacco leaf image is used as an impurity image; and impurity information is generated, and then the impurity information is transmitted to the ejection actuator;

[0028] The image processing module further includes: an image stitching sub-module; the image stitching sub-module continuously receives continuous multi-segment impurity images from the image recognition sub-module until no impurities are detected in the next image as a task completion flag;

[0029] The ejection actuator includes an ejection program and an ejection mechanism. After receiving the impurity information, the ejection program will control the ejection mechanism to eject the impurities.

[0030] Preferably, the image processing module is integrated and implemented in the image processing unit. The image processing unit identifies the image information and determines whether it is an impurity image, caches continuous multi-segment impurity images, stitches them into a complete impurity image through image stitching technology, and then transmits the impurity image to a gigabit Ethernet switch. The gigabit Ethernet switch uses gigabit Ethernet to transmit the impurity image to the host industrial control computer.

[0031] Beneficial effects:

[0032] The present invention proposes a method and system for real-time monitoring and early warning of key impurities based on optoelectronic impurity removal of cut tobacco. First, scan the tobacco leaves and collect images: In the optoelectronic impurity removal system of cut tobacco, use a line array or area array camera to scan the tobacco leaves and collect the image information of the tobacco leaves to obtain high-quality tobacco leaf images. Second, identify impurities: and determine whether there are impurities in the tobacco leaf image. If so, it is used as an impurity image. Third, splice the impurity images: Cache continuous multiple segments of impurity images, and splice the continuous multiple segments of impurity images into a complete impurity image through timestamps, and upload the complete impurity image to the upper industrial control computer using a switch and a gigabit network. Fourth, real-time monitoring and early warning of key impurities: The upper industrial control computer monitors the complete impurity image in real time, and the image is displayed on the statistical dashboard. Based on the AI classification algorithm, identify the key impurities and determine whether the impurity belongs to the key impurities. If it belongs to the key impurities, give an audible and visual alarm, and then further determine whether the key impurities appear continuously. If they appear continuously, the system issues a unique early warning prompt. Therefore, the present invention realizes the real-time monitoring and early warning of key impurities, solves the technical problem that the existing technology is in a black box state for impurity removal of tobacco leaves and cannot perform visual processing on impurity information, and discovers impurities in real time through the optoelectronic impurity removal system of cut tobacco combined with the AI classification algorithm of the upper computer for monitoring and early warning.

[0033] In the optoelectronic impurity removal system of cut tobacco, due to the very fast running speed of the material, no matter whether a line scan or area scan digital camera is used, only a small segment of the material image within the camera's field of view can be collected for analysis and processing at a certain moment. For some key impurities, their basic characteristics cannot be seen at all in this small segment of the image, and it is impossible to judge whether they are key impurities and what kind of key impurities they are based on this small segment of the image. Therefore, the present invention first proposes an image splicing method based on the optoelectronic impurity removal result. In step 3, continuous multiple segments of impurity images collected by the camera are spliced together to obtain a complete impurity image with a larger height for subsequent analysis and processing of key impurities. Description of the Drawings

[0034] Figure 1 It is a flow chart of a method for real-time monitoring and early warning of key impurities based on optoelectronic impurity removal of cut tobacco according to the present invention;

[0035] Figure 2 It is a structural schematic diagram of the optoelectronic impurity removal system based on the present invention;

[0036] Figure 3 It is an image processing architecture diagram of the present invention;

[0037] Figure 4 It is a working process schematic diagram of a method for real-time monitoring and early warning of key impurities based on optoelectronic impurity removal of cut tobacco in an embodiment;

[0038] Figure 5 Schematic diagram of an image stitching method proposed by the present invention.

[0039] Among them, 1 - linear array or area array camera, 2 - foreign object removal host computer software, 3 - gigabit network switch, 4 - FPGA image processing box, 5 - optoelectronic signal processing box, 6 - solenoid valve driver board, 7 - solenoid valve. Specific embodiments

[0040] The following further describes the specific embodiments of the present invention in combination with the accompanying drawings.

[0041] As Figure 1 shown, a real-time monitoring and early warning method for key sundries based on optoelectronic impurity removal of sliced tobacco includes the following steps:

[0042] Step 1: Scan the tobacco leaves and collect images: In the optoelectronic impurity removal system of sliced tobacco, use a linear array or area array camera to scan the tobacco leaves and collect the image information of the tobacco leaves;

[0043] Step 2: Identify sundries: And judge whether there are sundries in the tobacco leaf image. If so, it is used as a sundry image;

[0044] Step 3: Stitch the sundry images: Cache multiple consecutive segments of sundry images and record the timestamps until no sundries appear in the next image. Stop, and stitch the multiple consecutive segments of sundry images into a complete sundry image through the timestamps, and upload the complete sundry image to the upper industrial control computer through the switch and the gigabit network;

[0045] Step 4: Real-time monitoring and early warning of key sundries: The upper industrial control computer monitors the complete sundry image in real time. The image is displayed on the statistical dashboard, and based on the AI classification algorithm, the key sundries are identified, and it is judged whether the sundry belongs to the key sundries. If it belongs to the key sundries, an audible and visual alarm is issued.

[0046] Preferably, the identification of sundries in step 2 and the stitching of sundry images in step 3 are integrated and implemented in the image processing unit. The image processing unit identifies the image information and judges whether it is a sundry image, and caches multiple consecutive segments of sundry images, stitches them into a complete sundry image through image stitching technology, and then transmits the sundry image to the gigabit network switch. The gigabit network switch then uses the gigabit network to transmit the sundry image to the upper industrial control computer.

[0047] Preferably, the method for stitching sundry images in step 3 is as follows:

[0048] Step 3.1: Construct two image buffers, one is a defect image buffer and the other is a complete image buffer;

[0049] Step 3.2: After the image detection starts, the first detected foreign object image is stored in the defect image buffer. If the foreign object appears in at least one foreign object image, all the foreign object images with this foreign object are sequentially stored in the defect image buffer; continuously store the consecutive images of the complete foreign object, that is, until the last image is not a foreign object image, which means a stitching task is completed.

[0050] Step 3.3: According to the timestamps recorded in the foreign object images, all the foreign object images in the defect image buffer are stitched into a complete foreign object image and copied to the complete image buffer, and the complete foreign object image is uploaded to the host computer through the complete image buffer; and clear the defect image buffer until the next foreign object image enters, and then repeat Step 3.2.

[0051] Preferably, the method for identifying key foreign objects in Step 4 is as follows:

[0052] Use YOLOv5 in AI technology to build a model; the YOLOv5 built model is used to identify and classify foreign objects, and determine whether the foreign object belongs to a key foreign object. If it belongs to a key foreign object, the warning module controls the sound and light alarm to give an alarm. If the same type of key foreign object continuously appears within the same period, the warning module issues a key foreign object warning prompt, that is, the decibel of the sound emitted by the sound and light alarm increases, and the frequency of the flashing red light speeds up.

[0053] At the same time, the category, size, and quantity of the foreign object are displayed in real time on the key foreign object statistics dashboard, and it is judged whether the foreign object may be caused by a production equipment failure.

[0054] Preferably, the key foreign object analysis and warning steps can also: judge whether the key foreign object continuously appears within a period of time. If the foreign object continuously appears, the system will issue a key foreign object warning prompt.

[0055] A key foreign object warning system based on optoelectronic foreign object removal for sliced tobacco

[0056] It includes: a sliced tobacco optoelectronic foreign object removal system that identifies foreign object images and a host industrial computer that classifies, warns, and displays foreign object images.

[0057] The sliced tobacco optoelectronic foreign object removal system includes: an optical imaging module that captures tobacco leaf images, an image acquisition module that acquires tobacco leaf images, an image processing unit that identifies foreign objects and stitches foreign object images in the tobacco leaf images, and an ejection actuator that ejects foreign objects based on solenoid valves.

[0058] The host industrial computer includes: a key foreign object analysis and processing module that identifies key foreign objects based on an AI classification algorithm and a key foreign object statistics dashboard that displays key foreign object images.

[0059] Preferably, the optical imaging module consists of at least one linear array or area array camera, a lens, a light source, and a background;

[0060] The image acquisition module includes an image acquisition card, which acquires the tobacco leaf image information to generate a tobacco leaf image and further transmits the tobacco leaf image to the image recognition sub-module of the image processing module;

[0061] The image recognition sub-module processes the acquired image information to identify whether there are foreign matters in the tobacco leaf image. If there are foreign matters, the tobacco leaf image is taken as a foreign matter image; and foreign matter information is generated, and then the foreign matter information is transmitted to the rejection execution mechanism;

[0062] The image processing module further includes: an image stitching sub-module; the image stitching sub-module continuously receives continuous multi-segment foreign matter images from the image recognition sub-module until no foreign matters are detected in the next image as a task completion flag;

[0063] The rejection execution mechanism includes a rejection program and a rejection mechanism. After receiving the foreign matter information, the rejection program will control the rejection mechanism to remove the foreign matters.

[0064] Preferably, the image processing module is integrated and implemented in an image processing unit. The image processing unit recognizes the image information and determines whether it is a foreign matter image, caches continuous multi-segment foreign matter images, stitches them into a complete foreign matter image through image stitching technology, and then transmits the foreign matter image to a gigabit switch, and the gigabit switch uses gigabit network to transmit the foreign matter image to the upper industrial control computer.

[0065] The system structure of a real-time monitoring and early warning method for key foreign matters based on optoelectronic foreign matter removal of cut tobacco is as Figure 1 , and mainly includes units such as an optoelectronic foreign matter removal system, upper software, audible and visual alarms, and a statistical dashboard.

[0066] Among them, the structure of the optoelectronic foreign matter removal system is as Figure 2 , and mainly includes an optical imaging module, an image acquisition module, an image processing module, and a rejection execution mechanism;

[0067] The optical imaging module is mainly composed of a camera, a lens, a light source background, and creates imaging conditions for image acquisition;

[0068] The image acquisition module mainly includes an image acquisition card, acquires image information, and further transmits the acquired information to the image processing module;

[0069] The image processing module mainly includes image processing algorithms, processes the acquired image information, identifies the foreign matter information in the tobacco leaves, and then transmits the foreign matter information to the rejection execution mechanism;

[0070] The rejection execution mechanism includes a rejection program and a rejection mechanism. After receiving the foreign matter information, the rejection program will control the rejection mechanism to remove the foreign matter.

[0071] The image processing architecture of this invention patent is as Figure 3 shown. The linear array / area array camera 1 is used to obtain the images of tobacco leaves and foreign matters. The acquired image data is transmitted to the image processing unit, i.e., the FPGA image processing box 4, through interfaces such as Camera Link, GigE Vision, USB, etc. It will be first stored in the internal memory (so that it can be quickly read and written in the subsequent processing process). The FPGA has dual-channel transmission (different from the traditional single-channel transmission). On the one hand, the FPGA image processing box 4 identifies the foreign matter images in the tobacco leaves through the graphics processing function (FPGA K7 UN). The identification of foreign matter images includes identifying the foreign matters in the images based on the color recognition algorithm and the AI classification learning algorithm, and outputting the foreign matter signal. The foreign matter signal is transmitted to the optoelectronic signal processing box 5 of the rejection mechanism through one of the channels (24-channel IO control interface). The optoelectronic signal processing box 5 then converts the optical signal into an electrical signal and transmits it to the solenoid valve drive board 6. The solenoid valve drive board 6 will control the solenoid valve 7 to remove the foreign matters in the tobacco leaves. On the other hand, the FPGA image processing box 4 splices the continuously buffered multi-segment foreign matter image information into a complete image through the image stitching technology, and then transmits the foreign matter picture to the gigabit Ethernet switch 3 through another channel. The gigabit Ethernet switch 3 uses the gigabit Ethernet to transmit the picture to the foreign matter rejection host computer software 2 of the optoelectronic foreign matter removal system.

[0072] The described image stitching method is to construct two image buffers inside the image processing unit of the optoelectronic foreign matter removal system, one is the defective image buffer and the other is the complete image buffer. After the image detection starts, the first foreign matter image of the optoelectronic foreign matter removal system is first stored in the defective image buffer. If the foreign matter appears in multiple images, then multiple foreign matter images are sequentially stored in the buffer. All foreign matter images are stored until an image is not a foreign matter image. At this time, all the foreign matter images in the buffer are stitched into a complete foreign matter image and copied to the complete image buffer, and the defective image buffer is cleared to wait for the arrival of the next foreign matter image.

[0073] The present invention uses artificial intelligence technology to analyze and process key debris images. Based on AI technology, a mature YOLOv5 model is constructed. The AI model is used to identify debris, obtain the classification of the debris, determine whether the debris belongs to key debris. If it belongs to key debris, the size and attributes of the debris are obtained, and it is determined whether the debris may be caused by a production equipment failure. If the key debris analysis and processing module determines that the debris image is key debris, the warning module controls the sound and light alarm to issue a warning. If the same type of key debris continuously appears within the same time period, the warning module issues a key debris warning prompt, that is, the sound decibel emitted by the sound and light alarm increases, and the flashing frequency of the red light accelerates. The statistical dashboard displays the information of the key debris that appears in real time, including size, category, and quantity.

[0074] The working process of a real-time monitoring and warning method for key debris based on sheet tobacco photoelectric impurity removal of the present invention is as Figure 4 shown. First, a camera of the photoelectric impurity removal system is used to scan to obtain continuous multi-segment debris images, and then the continuous multi-segment debris images are stitched into a complete debris image through an image stitching method;

[0075] The image stitching method is as Figure 5 follows. Two image buffers are constructed inside the image processing module of the photoelectric impurity removal system, one is a defect image buffer and the other is a complete image buffer;

[0076] After the image detection starts, the first debris image of the photoelectric impurity removal system is first stored in the defect image buffer. If the next image of this debris image is also a debris image, then the next debris image is continuously stored in the buffer in sequence; all continuous debris images are stored until an image is not a debris image. At this time, all the debris images in the buffer are stitched into a complete debris image and copied to the complete image buffer, and the defect image buffer is emptied to wait for the arrival of the next debris image.

[0077] After the photoelectric impurity removal system obtains a complete debris image, it uploads it to the upper software through the image processing module in the photoelectric impurity removal system. The key debris analysis and processing module in the upper software determines whether the debris belongs to key debris. If it belongs to key debris, the system issues a sound and light alarm, and then further determines whether the key debris appears continuously. If it appears continuously, the system issues a specific warning prompt.

[0078] The statistical dashboard in the upper software will count the size, category, quantity, etc. of the key debris.

[0079] Those of ordinary skill in the art can realize that the units and method steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the embodiments of the present invention.

Claims

1. A real-time monitoring and early warning method for key debris based on photoelectric impurity removal of cigarette pieces, characterized in that: The steps include: Step 1: Scan tobacco leaves and collect images: In the tobacco leaf photoelectric impurity removal system, a linear array or area array camera is used to scan tobacco leaves and collect tobacco leaf image information; Step 2: Identify debris: determine whether there is debris in the tobacco leaf image, and if so, use it as a debris image; Step 3: Miscellaneous image stitching: Cache multiple consecutive segments of miscellaneous images and record timestamps until the next image does not contain miscellaneous objects, stitch the multiple consecutive segments of miscellaneous images into a complete miscellaneous image by timestamps, and upload the complete miscellaneous image to the upper industrial computer using a switch and a gigabit network; Step 4: Real-time monitoring and early warning of key debris: The upper industrial computer monitors the complete debris image in real time, and the image is displayed on the statistical dashboard. Based on the AI ​​classification algorithm, the key debris is identified and judged whether the debris is a key debris. If it is a key debris, an audible and visual alarm is issued; The method for splicing the debris images in step 3 is: Step 3.1: Construct two image buffers, one is a defect image buffer and the other is a complete image buffer; Step 3.2: After the image detection starts, the first identified debris image is first stored in the defect image buffer. If the debris appears in at least one debris image, all debris images with the debris are stored in the defect image buffer in sequence. The complete continuous images of the debris are stored until the last image is not a debris image, which is considered as a stitching task. Step 3.3: According to the timestamp recorded in the debris image, all debris images in the defect image buffer are stitched into a complete debris image and copied to the complete image buffer, and the complete debris image is uploaded to the host computer through the complete image buffer; and the defect image buffer is cleared until the next debris image enters and step 3.2 is repeated.

2. According to claim 1, a method for real-time monitoring and early warning of key debris based on photoelectric impurity removal of cigarette pieces is characterized by: The step 2 of identifying debris and the step 3 of splicing the debris images are integrated and implemented in an image processing unit. The image processing unit identifies the image information and determines whether it is a debris image, and caches multiple consecutive segments of debris images, splicing them into a complete debris image through image splicing technology, and then transmits the debris image to a gigabit network switch, which then uses the gigabit network to transmit the debris image to a host industrial computer.

3. The method for real-time monitoring and early warning of key debris based on photoelectric impurity removal of cigarette pieces according to claim 1 is characterized by: The method for identifying key debris in step 4 is: The model is constructed by using YOLOv5 in AI technology; the model identifies and classifies the debris and determines whether the debris is a key debris. If it is a key debris, the early warning module controls the sound and light alarm to issue an early warning. If the same type of key debris appears continuously in the same period, the early warning module issues a key debris early warning prompt, that is, the decibel of the sound emitted by the sound and light alarm increases, and the frequency of flashing red light increases; At the same time, the category, size and quantity of the debris are displayed in real time on the key debris statistics dashboard, and it is determined whether the debris is caused by production equipment failure.

4. The method for real-time monitoring and early warning of key debris based on photoelectric impurity removal of cigarette pieces according to claim 1 is characterized by: The key debris analysis and warning steps may also include: determining whether the key debris appears continuously within a period of time. If the debris appears continuously, the system will issue a key debris warning prompt.

5. A real-time monitoring and early warning system for key debris based on photoelectric impurity removal of cigarette sheets, comprising the method according to any one of claims 1 to 4, characterized in that: include: The photoelectric impurity removal system for cigarette pieces that identifies the impurity image and the upper industrial control computer that monitors the impurity image in real time and issues early warning for key impurities; The tobacco leaf photoelectric impurity removal system comprises: an optical imaging module for photographing tobacco leaf images, an image acquisition module for acquiring tobacco leaf images, an image processing unit for identifying impurities in tobacco leaf images and splicing impurity images, and a rejection actuator for rejecting impurities based on a solenoid valve; The upper industrial control computer includes: a key sundries analysis and processing module for identifying whether key sundries appear in an image based on an AI classification algorithm, and a key sundries statistics dashboard for real-time monitoring of key sundries images.

6. The key debris early warning system based on smoke sheet photoelectric debris removal according to claim 5 is characterized by: The optical imaging module is composed of at least one linear array or area array camera, a lens and a light source; The image acquisition module includes an image acquisition card, which acquires tobacco leaf image information to generate tobacco leaf images, and further transmits the tobacco leaf images to the image recognition submodule of the image processing module; The image recognition submodule processes the collected image information to identify whether there are foreign objects in the tobacco leaf image, and if there are foreign objects, the tobacco leaf image is used as a foreign object image; and generate debris information, which is then transmitted to a rejection actuator; The image processing module also includes: an image stitching submodule; The image stitching submodule continues to receive the continuous multiple segments of debris images from the image recognition submodule until the next image is detected to have no debris as a task completion mark; The rejection execution mechanism includes a rejection program and a rejection mechanism. After receiving the debris information, the rejection program controls the rejection mechanism to reject the debris.

7. The key debris early warning system based on smoke sheet photoelectric debris removal according to claim 6 is characterized in that: The image processing module is integrated in the image processing unit, which identifies the image information and determines whether it is a debris image, and caches multiple consecutive debris images, splices them into a complete debris image through image stitching technology, and then transmits the debris image to the Gigabit network switch, which then uses the Gigabit network to transmit the debris image to the upper industrial computer.

Images