Quality grading detection method and system for reconstituted tobacco papermaking section and storage medium

Abstract

The invention provides a quality grading detection method and system for a reconstituted tobacco papermaking section and a storage medium. The quality grading detection method comprises the steps that S1, a visible light image and an infrared image of the reconstituted tobacco papermaking section are obtained; s2, performing defect positioning processing based on the visible light image and the infrared image to obtain a target defect image; and S3, performing defect judgment on the target defect image based on the semantic segmentation model, and quantitatively evaluating the quality of the reconstituted tobacco papermaking section. Visible light and infrared images are synchronously collected, multi-modal fusion and dynamic correction are carried out in combination with humidity sensor data, and a high-resolution fusion image covering full breadth is generated; and then an improved ESOD target detection model is used to realize rapid identification and accurate positioning of multiple types of defects. According to the method, full-process automatic detection from defect screening and positioning to accurate quantitative analysis is realized, the manual intervention requirement is remarkably reduced, and the detection precision and real-time performance are improved.

Description

Technical Field

[0001] This application belongs to the field of tobacco leaf testing, specifically relating to a quality grading and testing method, system, and storage medium for reconstituted tobacco leaf processing. Background Technology

[0002] Reconstituted tobacco, as a crucial raw material in the tobacco industry, has a decisive impact on the sensory quality and stability of cigarette products due to its uniformity of quality. The papermaking process is the core of reconstituted tobacco production using the papermaking method. During this process, the pulp is dehydrated and shaped on a forming wire to form a continuous substrate. Tobacco leaves are highly susceptible to various defects during production, such as cracks, holes, black spots, dirt, and wrinkles. These defects not only affect the physical strength and processing performance of the reconstituted tobacco but also lead to uneven absorption of the coating solution, ultimately significantly reducing the intrinsic quality and utilization rate of the finished product.

[0003] Currently, quality inspection in the reconstituted tobacco leaf processing section mainly relies on two methods. First, manual inspection, which depends on quality inspectors visually observing and judging under standard light sources. This method has inherent drawbacks such as high subjectivity, low efficiency, and a tendency to miss defects, and it cannot provide real-time feedback on quality fluctuations during production. Second, traditional online visual inspection, which uses a single type of camera (such as a standard line scan camera) in its visual inspection system. However, this method has significant limitations. For example, its detection capabilities are limited; ordinary visible light cameras struggle to effectively identify defects with low contrast (black spots, dirt), and it's difficult to balance efficiency and accuracy, making it hard to create detailed images of suspected defect areas to confirm the defect type. Furthermore, there is a lack of collaborative mechanisms; existing camera systems are often simply arranged, operating independently, without effective data collaboration, failing to achieve the functional division of "preliminary screening" and "precise location." Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a quality grading and detection method, system and storage medium for the reconstituted tobacco leaf processing section, which aims to solve the problems of low efficiency, insufficient accuracy and difficulty in achieving defect grading and evaluation in the existing reconstituted tobacco leaf production process.

[0005] To solve the above-mentioned technical problems, this application adopts the following technical solution:

[0006] In a first aspect, this application provides a quality grading and detection method for a reconstituted tobacco processing section. The quality grading and detection method includes: step S1: acquiring visible light images and infrared images of the reconstituted tobacco processing section; step S2: performing defect localization processing based on the visible light images and infrared images to obtain target defect images; step S3: determining defects in the target defect images based on a semantic segmentation model to quantitatively evaluate the quality of the reconstituted tobacco processing section.

[0007] Further, step S1 includes: Step S11: synchronously scanning the reconstituted tobacco leaves using a dual-line array camera to obtain adjacent and continuous first and second visible light images; Step S12: stitching the first and second visible light images together to obtain a visible light image, which is a full-width visible light panoramic image; Step S13: acquiring images of the reconstituted tobacco leaf processing section using an infrared camera to obtain an initial infrared image; Step S14: performing spatial registration and grayscale normalization processing on the initial infrared image to obtain an infrared image.

[0008] Furthermore, step S1 includes: step S15: acquiring ambient humidity data; step S16: adaptively adjusting the brightness and contrast of the visible light image and the infrared image based on the ambient humidity data.

[0009] Further, step S2 includes: step S21: performing multimodal fusion of visible light image and infrared image to obtain multimodal fused image of reconstituted tobacco leaf processing section; step S22: inputting multimodal fused image into target detection model to obtain defect location information and defect size information, the target detection model being ESOD target detection model; step S23: obtaining target defect image based on defect location information and defect size information.

[0010] Furthermore, step S23 includes: step S231: moving the area array camera to the center of the defect based on the defect location information; step S232: adjusting the camera settings parameters of the area array camera to take a picture based on the defect size information to obtain the target defect image.

[0011] Furthermore, step S3 includes: step S31: inputting the target defect image into the semantic segmentation model, which is the U-Net++ semantic segmentation model; step S32: the semantic segmentation model performs defect region segmentation and classification processing on the target defect image and outputs segmentation mask information; step S33: judging the defects in the target defect image based on the segmentation mask information and quantitatively evaluating the quality of the reconstituted tobacco leaf processing section.

[0012] Further, step S33 includes: step S331: based on the segmentation mask information, calculate the defect geometric feature parameters in the target defect image, including area, perimeter, aspect ratio, and contour shape; step S332: extract defect texture statistical features according to the defect geometric feature parameters, including energy, entropy, and contrast; step S333: classify the defect texture statistical features according to a preset grading standard and / or clustering algorithm to obtain the quality assessment results of the reconstituted tobacco leaf processing section.

[0013] Secondly, this application provides a quality grading and detection system for the reconstituted tobacco leaf processing section. The quality grading and detection system includes an acquisition module, a defect location module, and a defect determination module. The acquisition module is used to acquire visible light images and infrared images of the reconstituted tobacco leaf processing section. The defect location module is used to perform defect location processing based on the visible light images and infrared images to obtain target defect images. The defect determination module is used to determine the defects in the target defect images based on a semantic segmentation model to quantitatively evaluate the quality of the reconstituted tobacco leaf processing section.

[0014] Thirdly, this application provides a computer-readable storage medium storing a computer program / instructions thereon, which, when executed by a processor, implements the steps of the above-mentioned quality grading and detection method for the reconstituted tobacco leaf papermaking section.

[0015] Fourthly, this application provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the above-mentioned quality grading and detection method for the reconstituted tobacco leaf papermaking section.

[0016] As can be seen from the above technical solution, the advantages and positive effects of the quality grading and detection method, system, and storage medium for the reconstituted tobacco leaf processing section proposed in this application are as follows:

[0017] (1) This invention uses a dual-line array camera and an infrared camera to achieve high-speed, wide-area, and equally spaced visible light and infrared images to be acquired simultaneously. It also combines humidity sensor data to perform multimodal fusion and dynamic correction to generate a high-resolution fused image covering the entire width. Subsequently, an improved ESOD small target detection model is used to achieve rapid identification and accurate localization of various types of defects. The detected defect areas are then input into an improved U-Net++ semantic segmentation model to complete pixel-level classification and mask extraction of the defect areas.

[0018] This method achieves fully automated detection from defect screening and location to precise quantitative analysis, significantly reducing the need for manual intervention, improving detection accuracy and real-time performance, and is suitable for high-speed continuous production processes in the reconstituted tobacco leaf processing section.

[0019] (2) This invention can output the type, geometric features, distribution range, and spatial location information of defects in real time during the production process, and perform dynamic compensation in combination with humidity sensor data to achieve a comprehensive graded evaluation of the quality of the reconstituted tobacco leaf papermaking section. The detection results can provide reliable data support for closed-loop control of the production line, optimization of process parameters, and quality traceability, thereby improving the level of intelligence in the production process and product consistency. In addition, this method has good versatility and can be extended to the fields of intelligent detection and quality control of other wide-width continuous materials such as fabrics and paper. Attached Figure Description

[0020] The above description of this application and the following detailed embodiments will be better understood when read in conjunction with the accompanying drawings. It should be noted that the drawings are merely examples of the claimed technical solutions.

[0021] Figure 1 This is a three-dimensional structural diagram of the quality grading and testing device for the reconstituted tobacco leaf processing section of this application;

[0022] Figure 2 This is a flowchart of the quality grading and testing process for the reconstituted tobacco papermaking section of this application;

[0023] Figure 3 This is a diagram of the ESOD network structure in this application;

[0024] Figure 4 This is the U-Net++ network structure diagram of this application.

[0025] The reference numerals in the attached figures are explained as follows:

[0026] Conveyor belt: 1;

[0027] Cameras: 2;

[0028] First moving device: 3;

[0029] Second moving device: 4;

[0030] Humidity sensor: 5;

[0031] Line scan cameras: 6;

[0032] Infrared cameras: 7;

[0033] First sliding gantry frame: 8;

[0034] Second sliding gantry frame: 9;

[0035] Area scan camera: 91. Detailed Implementation

[0036] The detailed features and advantages of this application are described below in the specific embodiments. The content of this description is sufficient to enable any person skilled in the art to understand the technical content of this application and implement it accordingly. Based on the specification, claims and drawings disclosed in this specification, a person skilled in the art can easily understand the related objectives and advantages of this application.

[0037] The invention will now be described with reference to the accompanying drawings, in which similar reference numerals denote similar elements. While specific structures and arrangements are discussed, it should be understood that this is done merely for illustrative purposes. Those skilled in the art will recognize that other structures and arrangements can be used without departing from the spirit and scope of the invention. It will be apparent to those skilled in the art that the invention can also be used in a variety of other applications.

[0038] In this specification and claims, several terms will be used, and unless otherwise indicated, these terms will be defined to have the following meanings:

[0039] The singular forms “a” and “the” include their corresponding plural forms. “At least one” means one or more, and “more” means two or more. “At least one of the following” or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be expressed as: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0040] All figures used to represent component amounts, properties (e.g., molecular weight), reaction conditions, etc., should be considered to be modified in all cases by the terms "within the unavoidable margin of error" or "about". Therefore, the numerical values ​​set forth herein are approximate and may vary depending on the desired properties sought to be obtained by the present invention. The principle of equivalents, which is applied to a minimum and not intended to limit the scope of the claims, should be applied, for example, each value should be interpreted at least according to the specified significant digits and by applying conventional rounding techniques.

[0041] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0042] In the description of this embodiment, it should be noted that the terms "upper", "lower", "inner", "bottom", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product is usually placed during use. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0043] Unless otherwise indicated, the following abbreviations have the following meanings, and any other abbreviations used herein but not defined have their generally accepted standard meanings:

[0044] All other terms used herein that are specifically defined herein shall have the general meaning understood by one of ordinary skill in the art, in particular meaning that, upon reading the claims, specification and drawings of this patent, one can directly and without doubt determine how the technical solution of this patent can be implemented.

[0045] Even if there are incomplete descriptions, omissions, or ambiguities in the grammar, words, punctuation, graphics, symbols, etc. of the claims, specification, and drawings of this patent, a person skilled in the art can still arrive at the only correct understanding by reading the claims, specification, and drawings as a whole without extensive reasoning or experimentation, and effectively exclude various incorrect interpretations that are not aimed at achieving the purpose of this patent.

[0046] Those skilled in the art would first choose to read the claims, specification, and drawings of this patent to reasonably interpret the terms; secondly, they would choose to refer to the relevant definitions in other documents published by the applicant before the filing date to reasonably interpret the terms; thirdly, they would choose the references cited in this patent to reasonably interpret the terms; and finally, they would choose to combine the technical dictionaries, technical manuals, reference books, textbooks, national or industry technical standards, etc., commonly used by those skilled in the art to reasonably interpret the terms.

[0047] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0048] Please refer to Figure 1 A quality grading and detection device for the papermaking process of reconstituted tobacco leaves is provided. The quality grading and detection device includes a conveyor belt 1, a camera 2, a first moving device 3, a second moving device 4, a humidity sensor 5, a line array camera 6, an infrared camera 7, a first sliding gantry 8, and a second sliding gantry 9.

[0049] Conveyor belt 1 is used to carry and transport the reconstituted tobacco raw materials to be tested, so as to achieve stable and continuous movement of the tested sample within the testing area.

[0050] Camera 2 is installed at the front end of conveyor belt 1 to monitor the material running status in real time and record the overall picture.

[0051] The first sliding gantry 8 and the second sliding gantry 9 are respectively positioned above and below the path of the conveyor belt 1. Both the first sliding gantry 8 and the second sliding gantry 9 are equipped with a first moving device 3 and a second moving device 4. The first moving device 3 is positioned above the conveyor belt 1, and the second moving device is positioned on both sides of the conveyor belt 1. The first moving device 3 and the second moving device 4 can be used to support and adjust the position of the imaging device above the sliding gantry.

[0052] An infrared camera 7 is installed on the first moving device 3 of the first sliding gantry frame 8, and a line scan camera 6 is installed on the upper crossbeam of the second moving device 4 of the first sliding gantry frame 8. The line scan camera 6 is used to acquire visible light images, and the infrared camera 7 is used to acquire infrared thermal images.

[0053] The first moving device 3 of the second sliding gantry 9 is equipped with an area array camera 91, which can directly capture images of the front of the reconstituted tobacco raw material.

[0054] Humidity sensor 5 is installed on the side of conveyor belt 1 to synchronously collect ambient humidity, providing auxiliary data support for multimodal fusion and defect feature correction. The two are arranged in parallel in space and the images are aligned through synchronous control, thereby obtaining surface texture and internal thermal feature information of reconstituted tobacco leaves.

[0055] The device described above in this invention, through multi-sensor collaborative design, can achieve synchronous acquisition and multi-modal fusion of visible light and infrared images, providing high-quality data input for subsequent defect identification and classification, and can perform dynamic compensation and correction based on humidity information, thereby achieving high-precision online monitoring of typical defects in the reconstituted tobacco leaf processing section.

[0056] Based on the same inventive concept, please refer to Figure 2 This application also provides a quality grading and testing method for the reconstituted tobacco leaf processing section, which can be applied to the above-mentioned quality grading and testing device. The specific steps of the quality grading and testing method are as follows:

[0057] Step S1: Acquire visible light and infrared images of the reconstituted tobacco leaf processing section.

[0058] Step S1 includes:

[0059] Step S11: The reconstituted tobacco leaves are scanned synchronously by the dual-line array camera 6 to obtain adjacent and continuous first and second visible light images.

[0060] Step S12: Stitch the first visible light image and the second visible light image together to obtain a visible light image, which is a full-width visible light panoramic image.

[0061] Step S13: The infrared camera 7 acquires images of the reconstituted tobacco leaf processing section to obtain an initial infrared image.

[0062] Step S14: Perform spatial registration and grayscale normalization on the initial infrared image to obtain the infrared image.

[0063] Step S15: Obtain ambient humidity data.

[0064] Step S16: Adaptively adjust the brightness and contrast of the visible light image and infrared image based on the ambient humidity data.

[0065] Specifically, a dual-line array camera 6 installed on the first sliding gantry 8 is used to simultaneously scan the high-speed moving reconstituted tobacco leaves and collect high-resolution visible light images of a portion of the width. At the same time, an infrared camera 7 installed on the same gantry is used to collect infrared images of the tobacco leaves and obtain environmental humidity data monitored in real time by a humidity sensor 5 located near the conveyor belt 1.

[0066] The first sliding gantry 8 has sliders at both ends that cooperate with guide rails set on both sides of the transmission unit. The extension direction of the guide rails is parallel to the transmission direction of the reconstituted tobacco leaves. The first sliding gantry 8 is driven and controlled by a servo motor, which controls the first sliding gantry 8 to move along the guide rail to a preset detection position.

[0067] On the crossbeam of the first sliding gantry 8, a dual line array camera 6 and an infrared camera 7 are installed via transverse guide rails. The line array camera 6 is mainly used to detect all defects, while the infrared camera 7 assists the line array camera 6 in detecting thickness-related defects (cracks, holes, wrinkles).

[0068] The tilt and yaw angles of the infrared camera 7 are adjusted by the gimbal to align its field of view with the detection area of ​​the line scan camera 6. The humidity sensor 5 collects ambient humidity in real time and inputs it into the control system for subsequent multimodal image dynamic correction and model feature compensation.

[0069] The adjacent images acquired by the two line scan cameras 6 are registered and stitched together according to the overlapping imaging area to obtain a full-width visible light panoramic image. The infrared image acquired by the infrared camera 7 is subjected to grayscale normalization and spatial registration processing to make it completely aligned with the panoramic visible light image in space.

[0070] Step S2: Perform defect localization processing based on visible light and infrared images to obtain the target defect image.

[0071] Step S2 includes:

[0072] Step S21: Perform multimodal fusion of the visible light image and the infrared image to obtain a multimodal fused image of the reconstituted tobacco leaf processing section.

[0073] A multi-modal image fusion algorithm employing multi-scale wavelet transform is used to complementarily fuse infrared and visible light information, enhancing the infrared response to thickness variations while preserving visible light detail and texture information. Data from a humidity sensor is then used to adaptively adjust the brightness and contrast of the fused image, eliminating the impact of humidity changes on imaging features.

[0074] Step S22: Input the multimodal fusion image into the target detection model to obtain defect location information and defect size information. The target detection model is the ESOD target detection model.

[0075] Please refer to Figure 3 This application identifies the category of defects using the ESOD model and outputs the pixel coordinates of their bounding boxes in the panoramic image.

[0076] The ESOD model first receives synchronized images from two sets of linear scan cameras, and then merges them into a single continuous input image using an image stitching algorithm based on feature matching or geometric correction to ensure the structural integrity of the wide tobacco surface.

[0077] The backbone network (Stem) of the model utilizes multi-layer convolutions, downsampling structures, and the C3k2 module evolved from the CSP structure to extract multi-scale features from images, thereby capturing the semantic expression of tobacco defects and providing a sufficient representational foundation for subsequent feature enhancement and detection. To further alleviate the characteristics of tobacco defects, such as small size, weak contrast, and sparse distribution, ESOD introduces an Obj Seeker module, which gives greater attention to potential small target regions through saliency enhancement and spatial response amplification mechanisms, enabling the network to strengthen its ability to perceive weak texture anomalies in the early stages.

[0078] Under the constraints of computational characteristics of wide-area, high-resolution images, the model employs an adaptive slicing module (Ada Slicer) to dynamically divide the feature map according to the feature distribution, representing it as a series of local regions (patches). This significantly reduces computational costs while maintaining cross-regional consistency, enabling the model to meet the high-speed inference requirements of industrial production lines while maintaining high accuracy.

[0079] Subsequently, the Neck feature fusion module integrates representations at different scales through operations such as upsampling, downsampling, and feature concatenation, and introduces structures such as C3k2 and A2C2f to achieve cross-layer information interaction, enabling high-resolution detailed information and low-resolution deep semantics to be represented in a coordinated manner, thereby enhancing the robustness of the model in multi-scale defect environments.

[0080] Finally, the Sparse Head performs defect classification and location regression based on the fused multi-scale features, and uses sparse attention and candidate point screening mechanisms to adapt to the sparse and random distribution of tobacco defects, outputting the category, confidence level and spatial coordinates of each defect, thus achieving high-precision detection of various defects on the tobacco surface.

[0081] Through the comprehensive design of strategies such as saliency guidance, dynamic slicing, multi-scale interaction and sparse prediction, the ESOD model can achieve efficient perception and accurate localization of minute tobacco defects in industrial environments with complex backgrounds, diverse textures and high real-time requirements.

[0082] Step S23: Obtain the target defect image based on the defect location information and defect size information.

[0083] Step S23 includes:

[0084] Step S231: Based on the defect location information, move the area array camera 91 to capture images at the center of the defect.

[0085] Step S232: Based on the defect size information, adjust the camera settings of the area array camera 91 to take a picture and obtain the target defect image.

[0086] Based on the defect pixel coordinate information obtained from the ESOD model, its spatial position in the world coordinate system is calculated, and the second sliding gantry 9 is controlled to drive the area array camera 91 on it to move directly above the spatial position.

[0087] The camera's focal length and height are dynamically adjusted based on the defect distribution range information, and a strobe light source is simultaneously triggered to acquire high-definition images. At the same time, the ambient humidity value during acquisition is recorded and stored in the image metadata.

[0088] Step S3: Defect determination is performed on the target defect image based on the semantic segmentation model to quantitatively evaluate the quality of the reconstituted tobacco leaf processing section.

[0089] Step S3 includes:

[0090] Step S31: Input the target defect image into the semantic segmentation model, which is the U-Net++ semantic segmentation model.

[0091] Step S32: The semantic segmentation model performs defect region segmentation and classification processing on the target defect image, and outputs segmentation mask information.

[0092] It is understandable that the target defect image (high-resolution close-up image of the defect) is input into the pre-trained U-Net++ semantic segmentation model, and the model performs pixel-level defect region segmentation and outputs accurate mask and category information.

[0093] Please refer to Figure 4 In the input image, the target regions (such as defects, objects, etc.) are marked with yellow boxes, and the network needs to segment or detect these regions.

[0094] In the backbone network, the resolution of the feature maps is gradually reduced by downsampling, while the number of channels is increased to extract high-level semantic features.

[0095] X 0,0 These represent feature maps at different levels. The deeper the level, the richer the semantic information, but the lower the spatial resolution.

[0096] The dashed arrow skip connections fuse shallow (high-resolution) features with deep (high-semantic) features, compensating for the spatial information lost during upsampling.

[0097] Upsampling gradually restores the feature map resolution and combines it with shallow features from skip connections to generate refined segmentation results.

[0098] The loss function (L) can be used to supervise network training, compare the output results with the ground truth, and optimize model parameters.

[0099] The model's final output is pixel-level segmentation results (such as red and blue regions), which may correspond to different categories or instances.

[0100] It is understandable that the U-Net++ semantic segmentation model extracts high-level semantic features through downsampling, restores resolution through upsampling, and uses skip connections to fuse multi-scale information to improve segmentation accuracy.

[0101] The U-Net++ model itself includes an encoder, a decoder, a nested skip connection module, and an attention fusion module.

[0102] The encoder is used to extract multi-scale semantic features, and the decoder gradually restores the spatial resolution through upsampling.

[0103] Nested skip connection modules introduce dense feature fusion between different layers to enhance boundary continuity and detail fidelity.

[0104] The attention fusion module combines channel attention and spatial attention mechanisms to adaptively adjust feature weights and improve the segmentation accuracy of minor defects.

[0105] The model employs depthwise separable convolution and batch normalization to achieve a balance between detection accuracy and computational efficiency.

[0106] Step S33: Defect judgment is performed on the target defect image based on the segmentation mask information, and the quality of the reconstituted tobacco leaf processing section is quantitatively evaluated.

[0107] Step S33 includes:

[0108] Step S331: Based on the segmentation mask information, calculate the geometric feature parameters of the defect in the target defect image. The geometric feature parameters of the defect include area, perimeter, aspect ratio and contour shape.

[0109] Step S332: Extract the defect texture statistical features based on the defect geometric feature parameters. The defect texture statistical features include energy, entropy, and contrast.

[0110] Step S333: Classify the statistical features of defect textures according to preset grading standards and / or clustering algorithms to obtain the quality assessment results of the reconstituted tobacco leaf processing section.

[0111] Based on segmentation masking, the key geometric feature parameters of defects are accurately quantified (including morphological indicators such as projected area, contour perimeter, aspect ratio and boundary complexity), and texture statistical features reflecting the internal heterogeneity of defects (such as energy distribution, information entropy, gray-scale contrast and local binary pattern features) are extracted.

[0112] By integrating machine vision inspection data with dynamic humidity data collected by the production line environmental monitoring system, an environmental compensation model based on multimodal data-driven model is constructed, which effectively eliminates the interference of environmental temperature and humidity fluctuations on defect characterization.

[0113] Based on this, the severity of defects is intelligently classified and determined according to industry grading standards or by using unsupervised clustering algorithms, generating a comprehensive analysis report that includes a visualized defect distribution heatmap and statistical reports, ultimately realizing intelligent quantitative assessment and process quality traceability of defects in the reconstituted tobacco leaf processing section.

[0114] Based on the same inventive concept, this application also provides a quality grading and detection system for the reconstituted tobacco leaf processing section, which includes an acquisition module, a defect location module, and a defect determination module.

[0115] The acquisition module is used to acquire visible light and infrared images of the reconstituted tobacco leaf processing section.

[0116] The defect localization module is used to perform defect localization processing based on visible light images and infrared images to obtain target defect images.

[0117] The defect determination module is used to determine the defects in the target defect image based on the semantic segmentation model, and to quantitatively evaluate the quality of the reconstituted tobacco leaf processing section.

[0118] Furthermore, the acquisition module is also used to synchronously scan the reconstituted tobacco leaves using the dual-line array camera 6 to obtain adjacent and continuous first and second visible light images; to stitch the first and second visible light images together to obtain a visible light image, which is a full-width visible light panoramic image; to acquire images of the reconstituted tobacco leaf processing section using the infrared camera 7 to obtain an initial infrared image; and to perform spatial registration and grayscale normalization processing on the initial infrared image to obtain an infrared image.

[0119] Furthermore, the acquisition module is also used to acquire ambient humidity data; and to adaptively adjust the brightness and contrast of visible light and infrared images based on the ambient humidity data.

[0120] Furthermore, the defect localization module is also used to perform multimodal fusion of visible light images and infrared images to obtain a multimodal fused image of the reconstituted tobacco leaf processing section; input the multimodal fused image into the target detection model to obtain defect location information and defect size information, the target detection model being the ESOD target detection model; and obtain the target defect image based on the defect location information and defect size information.

[0121] Furthermore, the defect location module is also used to move the area array camera 91 to capture images at the center of the defect based on the defect location information; and to adjust the camera setting parameters of the area array camera 91 to take pictures based on the defect size information, thereby obtaining an image of the target defect.

[0122] Furthermore, the defect determination module is also used to input the target defect image into the semantic segmentation model, which is the U-Net++ semantic segmentation model; the semantic segmentation model performs defect region segmentation and classification processing on the target defect image and outputs segmentation mask information; based on the segmentation mask information, the target defect image is determined to quantitatively evaluate the quality of the reconstituted tobacco leaf processing section.

[0123] Furthermore, the defect determination module is also used to calculate the defect geometric feature parameters in the target defect image based on the segmentation mask information. The defect geometric feature parameters include area, perimeter, aspect ratio, and contour shape. Based on the defect geometric feature parameters, the module extracts defect texture statistical features, which include energy, entropy, and contrast. The module then grades the defect texture statistical features according to a preset grading standard and / or clustering algorithm to obtain the quality assessment results of the reconstituted tobacco leaf processing section.

[0124] It is understood that the quality grading and testing system for the reconstituted tobacco papermaking section provided in this application corresponds to the quality grading and testing method for the reconstituted tobacco papermaking section provided in this application. For the sake of brevity, the same or similar parts can be referred to the content of the quality grading and testing method for the reconstituted tobacco papermaking section, and will not be repeated here.

[0125] The various modules in the quality grading and inspection system of the reconstituted tobacco leaf processing section described above can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the server in hardware form or independent of it, or stored in the server's memory in software form, so that the processor can call and execute the corresponding operations of each module. The processor can be a central processing unit (CPU), a microprocessor, a microcontroller, etc.

[0126] The aforementioned quality grading and testing method and / or system for the reconstituted tobacco papermaking process can be implemented as a computer-readable instruction that can run on a computer system.

[0127] This application also provides a computer system including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the program, it implements the aforementioned quality grading and detection system for the reconstituted tobacco leaf processing section.

[0128] The computer system can be a server. The computer system includes a processor, non-volatile storage medium, internal memory, input devices, a display screen, and a network interface connected via a system bus. The non-volatile storage medium of the computer system can store an operating system and computer-readable instructions. When these computer-readable instructions are executed, they can cause the processor to execute a quality grading and detection method for a reconstituted tobacco leaf processing section according to various embodiments of this application. The specific implementation process of this method can be found in [reference needed]. Figure 2 The specific details will not be elaborated here.

[0129] The processor of this computer system provides computing and control capabilities, supporting the operation of the entire system. The internal memory stores computer-readable instructions, which, when executed by the processor, enable the processor to perform a quality grading and detection method for the reconstituted tobacco leaf processing section. The computer system's input devices are used for inputting various parameters, its display screen is used for display, and its network interface is used for network communication.

[0130] Based on the same inventive concept, this application provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps in the above-mentioned quality grading and detection method for the reconstituted tobacco leaf processing section.

[0131] The memory in the embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0132] The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0133] Volatile memory can be random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct memory bus RAM (DR RAM).

[0134] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs.

[0135] When computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means.

[0136] Computer-readable storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives (SSDs).

[0137] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0138] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0139] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0140] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0141] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0142] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0143] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0144] In this specification, references to "an embodiment" or "a specific implementation" mean that a particular feature, structure, or characteristic described in connection with that embodiment / specific implementation is included in at least one embodiment / specific implementation of the invention. Therefore, the phrase "in one embodiment / specific implementation" appearing in various places in this specification does not necessarily refer to the same embodiment / setting, but rather to potentially different embodiments. Furthermore, specific features, structures, or characteristics may be combined in one or more embodiments / settings in any suitable manner, as will be apparent to those skilled in the art from this disclosure.

[0145] Similarly, it should be understood that in the above description of exemplary embodiments / specific implementations of the present invention, various features of the present invention are sometimes combined together in a single embodiment / specific implementation or its figures and descriptions, with the aim of simplifying the disclosure and aiding in the understanding of one or more of the various aspects of the invention.

[0146] However, except for explicit instructions to the contrary or obvious technical contradictions or exclusions, the descriptive method of this patent should not be interpreted as reflecting an intention that the claimed features of the invention are more than those explicitly stated in each claim.

[0147] Conversely, the inventive aspect reflected in the claims lies in not all the features of a single of the foregoing disclosed embodiments / specific implementations.

[0148] Therefore, the claims following the detailed description are expressly incorporated herein, and each claim exists independently as a separate embodiment / specific implementation of the invention.

[0149] Furthermore, while some embodiments / specific implementations described herein include, but are not limited to, other features included in other embodiments / specific implementations, combinations of features from different embodiments / specific implementations are intended to be within the scope of the invention and form different embodiments / specific implementations, as will be understood by those skilled in the art. For example, in the following claims, embodiments / specific implementations of any claim can be used in any combination.

[0150] The terms and expressions used in this specification are for illustrative purposes and not for limitation. In using these terms and expressions, it is not intended to exclude any equivalents of the features or portions thereof shown and described, but rather to recognize that various modifications may be possible within the scope of the invention.

[0151] Therefore, it should be understood that although the invention has been specifically disclosed through preferred embodiments, exemplary embodiments and optional features, those skilled in the art may take variations or modifications of the concepts disclosed herein, and such variations and modifications are therefore considered to be within the scope of the invention as defined by the appended claims.

[0152] The specific embodiments given in this specification are examples of useful implementations of the present invention. It will be apparent to those skilled in the art that the present invention can be implemented using many variations of the devices, device components, and method steps disclosed in this specification.

[0153] The foregoing description of specific embodiments fully discloses the general features of the present invention, enabling others to easily modify and / or adapt such specific embodiments for various applications by applying knowledge within the scope of the art, without conducting excessive experimentation and without departing from the general concept of the present invention.

[0154] Therefore, based on the teachings and guidance provided herein, it is intended that such modifications and alterations be included within the meaning and scope of equivalents of the disclosed embodiments. It should be understood that the wording or terminology used herein is for descriptive purposes and is not intended to be limiting; thus, the wording or terminology in this specification will be interpreted by those skilled in the art based on the foregoing teachings and guidance.

[0155] Furthermore, the scope of the invention should not be limited to any of the exemplary embodiments described above, but only to the appended claims and their equivalents.

Claims

1. A method for quality grading of a reconstituted tobacco leaf papermaking section, characterized by, The quality grading detection method comprises: Step S1: acquiring a visible light image and an infrared image of the reconstituted tobacco papermaking section; Step S2: performing defect positioning processing based on the visible light image and the infrared image to obtain a target defect image; Step S3: performing defect judgment on the target defect image based on a semantic segmentation model to quantitatively evaluate the quality of the reconstituted tobacco papermaking section.

2. The method of claim 1, wherein the quality grading of the reconstituted tobacco leaf papermaking section is characterized by, The step S1 comprises: Step S11: synchronously scanning the reconstituted tobacco according to a double linear array camera to obtain adjacent and continuous first and second visible light images; Step S12: splicing the first and second visible light images to obtain the visible light image, which is a full-width visible light panoramic image; Step S13: acquiring an initial infrared image of the reconstituted tobacco papermaking section according to an infrared camera; Step S14: performing spatial registration and gray scale normalization processing on the initial infrared image to obtain the infrared image.

3. The method of claim 1, wherein the quality grading of the reconstituted tobacco leaf papermaking section is characterized by, The step S1 comprises: Step S15: acquiring environmental humidity data; Step S16: adaptively adjusting the brightness and contrast of the visible light image and the infrared image according to the environmental humidity data.

4. The method of claim 1, wherein the quality grading of the reconstituted tobacco leaf papermaking section is characterized by, The step S2 comprises: Step S21: performing multi-modal fusion on the visible light image and the infrared image to obtain a multi-modal fusion image of the reconstituted tobacco papermaking section; Step S22: inputting the multi-modal fusion image into a target detection model to obtain defect position information and defect size information, the target detection model being an ESOD target detection model; Step S23: obtaining the target defect image according to the defect position information and the defect size information.

5. The method of claim 4, wherein the quality grading of the reconstituted tobacco leaf papermaking section is characterized by, The step S23 comprises: Step S231: moving a face array camera to the center position of the defect according to the defect position information; Step S232: adjusting camera setting parameters of the face array camera to take a picture according to the defect size information to obtain the target defect image.

6. The method of claim 1, wherein the quality grading of the reconstituted tobacco leaf papermaking section is characterized by, The step S3 comprises: Step S31: inputting the target defect image into the semantic segmentation model, the semantic segmentation model being a U-Net++ semantic segmentation model; Step S32: performing defect region segmentation and classification processing on the target defect image by the semantic segmentation model to output segmentation mask information; Step S33: performing defect judgment on the target defect image according to the segmentation mask information to quantitatively evaluate the quality of the reconstituted tobacco papermaking section.

7. The method of quality grading of a reconstituted tobacco leaf papermaking section according to claim 6, characterized in that, The step S33 comprises: Step S331: calculating defect geometric feature parameters in the target defect image based on the segmentation mask information, the defect geometric feature parameters including area, perimeter, aspect ratio and contour shape; Step S332: extracting defect texture statistical features according to the defect geometric feature parameters, the defect texture statistical features including energy, entropy and contrast; Step S333: grading the defect texture statistical features according to a preset grading standard and / or a clustering algorithm to obtain the quality evaluation result of the reconstituted tobacco papermaking section.

8. A quality grading detection system for a reconstituted tobacco leaf papermaking section, characterized by, The quality grading detection system comprises an acquisition module, a defect positioning module and a defect judgment module, The acquisition module is configured to acquire a visible light image and an infrared image of the reconstituted tobacco papermaking section; The defect positioning module is configured to perform defect positioning processing based on the visible light image and the infrared image to obtain a target defect image; The defect judgment module is configured to perform defect judgment on the target defect image based on a semantic segmentation model, and quantitatively evaluate the quality of the reconstituted tobacco papermaking section.

9. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to realize the steps of the quality grading detection method of the reconstituted tobacco papermaking section according to any one of claims 1-7.

10. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instruction is executed by the processor to realize the steps of the quality grading detection method of the reconstituted tobacco papermaking section according to any one of claims 1-7.

Images