Health care product production monitoring control method and system based on image recognition

Abstract

The invention relates to the technical field of computer vision, and discloses a health care product production monitoring control method and system based on image recognition, and the method comprises the following steps: S1, data collection: carrying out the real-time collection of a health care product through an industrial camera, an infrared imaging device and an environment sensor; s2, image processing: carrying out denoising, enhancement and edge detection on the acquired image data; s3, image recognition: analyzing the processed image by using a convolutional neural network (CNN); s4, defect detection and early warning are carried out, product quality is evaluated according to the identified defects, and a quality report is generated; s5, production control: automatically adjusting production process parameters according to a defect detection result; and S6, data storage and traceability: storing the collected production data, defect information and quality control parameters in a database. In the invention, through an image identification technology, defects of the health care products can be automatically identified and classified, and low efficiency and inconsistency of traditional manual detection are avoided.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and particularly to a health product production monitoring and control method and system based on image recognition. Background Art

[0002] With the rapid development of the health product industry, consumers have higher and higher requirements for the quality of health products, especially in the monitoring and control of product quality during the production process. Traditional quality control methods usually rely on manual inspection, but with the expansion of production scale and the improvement of production automation, manual inspection can no longer meet the requirements of efficient and accurate quality control. Therefore, how to achieve real-time monitoring and automated quality control during the production process has become an urgent problem to be solved in the health product production industry.

[0003] Most of the existing health product production monitoring methods rely on manual inspection or simple machine control, but the efficiency of manual inspection is low and the misjudgment rate is high: manual inspection not only requires a large amount of manpower, but is also easily affected by subjective factors. Especially during high-speed and large-batch production, manual quality inspection may not be able to detect all defects in time. Summary of the Invention

[0004] To make up for the above deficiencies, the present invention provides a health product production monitoring and control method and system based on image recognition, aiming to improve the problem that most of the existing health product production monitoring methods rely on manual inspection or simple machine control, and the efficiency of manual inspection is low and the misjudgment rate is high.

[0005] In the first aspect, the present invention provides the following technical solution. A health product production monitoring and control method based on image recognition includes the following steps:

[0006] S1. Data acquisition: Real-time acquisition of image data, temperature data and environmental parameters of the health product production line through industrial cameras, infrared imaging devices and environmental sensors;

[0007] S2. Image processing: Denoising, enhancing and edge detection are performed on the acquired image data to extract the appearance features, shape and size information of the health products;

[0008] S3. Image recognition: Use a convolutional neural network (CNN) to analyze the processed image, identify the defects in the health products, and classify the defects;

[0009] S4. Defect detection and warning: Evaluate the product quality according to the identified defects, generate a quality report, and automatically issue a warning according to the severity of the defects, triggering an automatic adjustment during the production process;

[0010] S5. Production control: Automatically adjust the production process parameters according to the results of defect detection, including temperature, humidity and production speed;

[0011] S6. Data storage and traceability, storing the collected production data, defect information, and quality control parameters in a database, and using blockchain technology to ensure data security and traceability.

[0012] Preferably, in the data collection step, the image data collected by the industrial camera is high-definition images, and the collection area of the industrial camera covers all links of the health product production line.

[0013] Preferably, the image processing specifically includes:

[0014] S201. Denoising processing, applying median filtering or Gaussian filtering to the image, where the median filtering window is 3x3 pixels or 5x5 pixels; Gaussian filtering uses a Gaussian kernel function with a standard deviation of 0.8 or 1.0 for smoothing processing;

[0015] S202. Image enhancement, using the adaptive histogram equalization algorithm to enhance the details of low-contrast image areas and increase the contrast by 50% to 200%;

[0016] S203. Edge detection, using the Canny edge detection algorithm with a minimum threshold of 100 and a maximum threshold of 200, or using the Sobel operator to extract edge information in the image.

[0017] Preferably, the image recognition specifically includes the following steps:

[0018] S301. Using a convolutional neural network CNN to extract low-level and high-level features in the image

[0019] In the convolutional layer of the convolutional neural network, multiple convolutional kernels are applied to extract local features, including edges, corners, and textures;

[0020] In the pooling layer, max pooling or average pooling operations are used, and the pooling window is 2x2 or 3x3, which is used to reduce the size of the feature map and retain key information;

[0021] The number of layers of the CNN network is 4 convolutional layers and 2 fully connected layers to extract image features of different scales and different levels;

[0022] S302. Based on the labeled image dataset, optimizing the weights of the convolutional neural network through the backpropagation algorithm and the gradient descent method

[0023] The training dataset contains at least 5000 images, each image contains a corresponding label, and each category contains at least 1000 images;

[0024] Training is carried out using batch gradient descent, with each batch containing 64 images. The Adam optimizer is used during training, and the learning rate is set to 0.001;

[0025] The CNN model is trained through 50 iterations, and the validation set is evaluated at each iteration, finally achieving a training accuracy exceeding 95%;

[0026] S303. Perform defect detection and classification on the input image through the trained convolutional neural network

[0027] Perform forward propagation on the input image. The CNN outputs the probability of each defect category. The Softmax function is used to normalize the output results, and the category probabilities are output. The category with the highest probability is selected as the final defect type;

[0028] The defect categories output by the CNN classifier include "crack", "deformation", "color difference", and "no defect". The probability value range for each category is from 0 to 1, and the sum of the probabilities is 1;

[0029] Precisely locate the defects in the image through the object detection algorithm:

[0030] Use the Faster R-CNN or YOLO algorithm to precisely locate the position of the defect by generating candidate boxes and further performing regression adjustment;

[0031] The position of each candidate box is precisely adjusted through region regression, making the positioning error less than 10%;

[0032] Output the position information of each defect and calibrate the coordinates of the box where the defect is located.

[0033] Preferably, in the defect detection and warning step, defect evaluation comprehensively evaluates different types of defects according to the fuzzy logic control FLC or support vector SVM method, generates a quality report, and automatically triggers a warning signal according to a preset threshold;

[0034] In the defect detection and warning step, the generated quality report includes not only defect information but also environmental data, production parameters, and adjustment history records of each production link for production management personnel to refer to;

[0035] The warning signal in the defect detection and warning step includes sound, visual, or email notifications.

[0036] Preferably, in the production control step, the adjustment of production process parameters includes automatically adjusting the temperature, humidity, and pressure parameters of production equipment, and precise control is carried out through the PID control algorithm or the adaptive control algorithm.

[0037] Preferably, in the data storage and traceability step, Ethereum or Hyperledger Fabric platform is adopted for the blockchain, which is responsible for the transparency, credibility, and immutability of the traceability data.

[0038] In a second aspect, the present invention provides the following technical solution. A health product production monitoring and control system based on image recognition includes:

[0039] A data acquisition module that collects image data, temperature data, and environmental parameters of the health product production line in real time through industrial cameras, infrared imaging devices, and environmental sensors;

[0040] An image processing module that performs denoising, enhancement, edge detection, and feature extraction on the collected image data to improve the accuracy of subsequent image analysis;

[0041] An image recognition module that uses a convolutional neural network (CNN) to perform in-depth analysis on the processed image data, identify defects of health products, and classify them;

[0042] A defect detection and warning module that, based on the identified defects, performs quality assessment and generates warning signals according to the severity of the defects, and automatically adjusts the production process;

[0043] A production control module that automatically adjusts production process parameters, including temperature, humidity, and production speed, according to the defect detection results;

[0044] A data storage and traceability module that stores production data, defect information, and quality control parameters in a database, and uses blockchain technology to be responsible for the security and traceability of the data.

[0045] In a third aspect, the invention provides the following technical solution. A computer device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned health product production monitoring and control method based on image recognition.

[0046] In a fourth aspect, the present invention provides the following technical solution. A readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the above-mentioned health product production monitoring and control method based on image recognition.

[0047] The present invention has the following beneficial effects:

[0048] 1. In the present invention, through image recognition technology, especially convolutional neural network (CNN), it is able to automatically identify and classify the defects of health products, avoiding the inefficiency and inconsistency of traditional manual inspection. The image recognition module can detect each product on the production line in real time, comprehensively evaluate the appearance, shape, color, etc., and promptly identify defects such as cracks, deformations, and color differences. This automated inspection not only greatly improves production efficiency but also reduces human oversights and enhances the accuracy of inspection.

[0049] 2. In the present invention, the defect detection and warning module can, based on the results of image recognition, evaluate the product quality in real time and generate warning signals. This mechanism can automatically trigger the adjustment of the parameters of production equipment (such as temperature, humidity, production speed, etc.) during the production process, optimize according to different defect situations, and ensure that the quality of each batch of health products meets the standards. This intelligent feedback control system reduces the dependence on manual intervention and improves the flexibility and response speed of the production line.

[0050] 3. In the present invention, through comprehensive image recognition and automated control, the need for manual inspection and intervention is reduced. This method realizes the automatic detection and classification of product defects through precise image recognition and deep learning algorithms, improves the consistency of quality control, ensures that each batch of products meets the quality standards, and reduces the quality fluctuations caused by human oversights.

[0051] 4. In the present invention, the data storage and traceability module ensures the security and immutability of production data through blockchain technology, enabling the production process of each batch of health products to be traced back to the source. All data such as production data, defect information, and production adjustment records can be generated and recorded in the blockchain, ensuring the transparency of product quality information and production history. This highly transparent traceability system enhances consumers' trust and also provides reliable data support for regulatory agencies, ensuring product safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is the flowchart of the health product production monitoring and control method based on image recognition proposed by the present invention;

[0053] Figure 2 is the flowchart of the image processing steps of the health product production monitoring and control method based on image recognition proposed by the present invention;

[0054] Figure 3 is the flowchart of the image recognition steps of the health product production monitoring and control method based on image recognition proposed by the present invention;

[0055] Figure 4 is the system architecture diagram of the health product production monitoring and control system based on image recognition proposed by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0056] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0057] Embodiment 1

[0058] Refer to Figures 1 - 3 , in the first embodiment of the present invention, the present invention provides a health product production monitoring and control method based on image recognition, including the following steps:

[0059] S1. Data acquisition: Real-time acquisition of image data, temperature data, and environmental parameters of the health product production line through industrial cameras, infrared imaging devices, and environmental sensors;

[0060] S2. Image processing: Denoising, enhancing, and edge detection of the acquired image data to extract the appearance features, shape, and size information of the health products;

[0061] S3. Image recognition: Analyzing the processed image using the convolutional neural network CNN to identify the defects in the health products and classify the defects;

[0062] S4. Defect detection and warning: Evaluating the product quality based on the identified defects, generating a quality report, and automatically issuing a warning according to the severity of the defects to trigger automatic adjustment during the production process;

[0063] S5. Production control: Automatically adjusting the production process parameters, including temperature, humidity, and production speed, according to the results of defect detection;

[0064] S6. Data storage and traceability: Storing the acquired production data, defect information, and quality control parameters in a database and ensuring the security and traceability of the data through blockchain technology.

[0065] Specifically, the health product production monitoring and control method based on image recognition can significantly improve production efficiency and reduce labor costs by realizing automated defect detection and classification. This method monitors the production process in real time, automatically identifies and classifies defects in health products, and generates warning signals immediately when problems are detected, triggering automatic adjustment of the production process to ensure the consistency and high standards of product quality. This process reduces manual intervention and improves the intelligence level of the production line through deep learning models and image recognition technologies. At the same time, blockchain technology is used to trace the production data to ensure the transparency and traceability of the production process of each batch of products, enhancing product safety and consumer trust. Therefore, the entire system not only improves the efficiency and reaction speed of the production line, but also ensures the quality stability and high transparency of each batch of products, with significant economic and social value.

[0066] In the data acquisition step, the image data collected by the industrial camera is high-definition, and the acquisition area of the industrial camera covers all aspects of the health product production line.

[0067] Specifically, on the health product production line, the data acquisition module is the starting point of the entire system, which is responsible for collecting image and environmental data from the production site. The data acquisition module includes the following equipment and technologies:

[0068] Industrial camera

[0069] The industrial camera is a key device in the data acquisition module for obtaining images of health products on the production line. According to the size, shape of the health product and the running speed of the production line, cameras with appropriate resolutions are selected (such as 1920x1080 or higher resolutions). These cameras usually have high frame rates and automatic adjustment functions, which can ensure the clarity and accuracy of images during the fast production process.

[0070] Working principle: The industrial camera captures the appearance of health products in real time through the camera and transmits the captured images to the data processing system. The installation position of the camera is adjusted according to the layout of the production line, and multiple cameras are usually installed at multiple important nodes (such as packaging, inspection and encapsulation, etc.).

[0071] Image acquisition settings: The camera is usually configured with a shutter synchronized with the light source to ensure the uniformity of illumination during image shooting and avoid image quality problems caused by changes in illumination.

[0072] Infrared imaging device

[0073] Used to detect the temperature distribution of health products during the production process, especially for health products that require precise temperature control (such as capsules, tablets, etc.), the infrared imaging device can capture the heat changes of products during the production process.

[0074] Working Principle: The infrared thermal imager generates a thermal map by capturing the infrared radiation signals on the product surface, showing the temperature changes. The device can precisely detect temperature anomalies by adjusting the sensitivity of the infrared camera.

[0075] Data Transmission: The acquired infrared thermal images are transmitted to the image processing unit through a dedicated interface.

[0076] Environmental Sensor

[0077] The environmental sensor is used to monitor various parameters of the production environment in real time, such as temperature, humidity, air pressure, etc. The data collected by the sensor can be subjected to correlation analysis in the subsequent quality control process to help identify environmental factors that may affect product quality.

[0078] Device Selection: Sensors with high precision and fast response speed (such as DHT11 and DHT22 temperature and humidity sensors, pressure sensors, etc.) are adopted to ensure real-time performance and accuracy.

[0079] Working Principle: The environmental sensor sends data to the central data processing platform through the Modbus protocol or a wireless network (such as LoRa). The real-time data of the sensor helps managers adjust production conditions to avoid affecting the quality of health products due to improper environment.

[0080] Image processing specifically includes:

[0081] S201, Denoising Processing: Apply median filtering or Gaussian filtering to the image. The median filtering window is 3x3 pixels or 5x5 pixels; Gaussian filtering uses a Gaussian kernel function with a standard deviation of 0.8 or 1.0 for smoothing.

[0082] S202, Image Enhancement: Use the adaptive histogram equalization algorithm to enhance the details of low-contrast image areas, increasing the contrast by 50% to 200%.

[0083] S203, Edge Detection: Adopt the Canny edge detection algorithm, set the minimum threshold to 100 and the maximum threshold to 200, or use the Sobel operator to extract the edge information in the image.

[0084] Specifically, the image processing module is responsible for performing various preprocessing and enhancement operations on the acquired images to improve the image quality and ensure that the subsequent image recognition module can accurately extract effective information.

[0085] During the image acquisition process, the image may be disturbed by noise, including granular interference or image blurring in the image. To solve this problem, denoising algorithms are used to reduce unnecessary image noise.

[0086] The purpose of image enhancement is to improve the contrast, brightness, etc. of the image, so as to more clearly identify the appearance details of health products. Especially in low-light environments, image enhancement helps to brighten the dark areas and highlight the defect details.

[0087] Edge detection is an important step in image processing. It helps the system identify the object contour by extracting the edge information in the image, so as to detect the features such as the shape and size of health products.

[0088] Image recognition specifically includes the following steps:

[0089] S301. Use the convolutional neural network CNN to extract low-level and high-level features in the image

[0090] In the convolutional layer of the convolutional neural network, multiple convolutional kernels are applied to extract local features, including edges, corners, and textures;

[0091] In the pooling layer, use the max pooling or average pooling operation, and the pooling window is 2x2 or 3x3, which is used to reduce the size of the feature map and retain key information;

[0092] The number of layers of the CNN network is 4 convolutional layers and 2 fully connected layers to extract image features of different scales and different levels;

[0093] S302. Based on the labeled image dataset, optimize the weights of the convolutional neural network through the backpropagation algorithm and the gradient descent method

[0094] The training dataset contains at least 5000 images, and each image contains a corresponding label, where each category contains at least 1000 images;

[0095] Use the batch gradient descent method for training. Each batch contains 64 images. The Adam optimizer is used during training, and the learning rate is set to 0.001;

[0096] Train the CNN model through 50 iterations, and evaluate the validation set at each iteration. Finally, the training accuracy exceeds 95%;

[0097] S303. Perform defect detection and classification on the input image through the trained convolutional neural network

[0098] Perform forward propagation on the input image. The CNN outputs the probability of each defect category. Use the Softmax function to normalize the output result, output the category probability, and select the category with the highest probability as the final defect type;

[0099] The defect categories output by the CNN classifier include "crack", "deformation", "color difference", and "no defect". The probability value range of each category is from 0 to 1, and the sum of probabilities is 1;

[0100] Precisely locate the defects in the image through an object detection algorithm:

[0101] Use the Faster R-CNN or YOLO algorithm to precisely locate the position of the defect by generating candidate boxes and further performing regression adjustment;

[0102] The position of each candidate box is precisely adjusted through regional regression, making the positioning error less than 10%;

[0103] Output the position information of each defect and calibrate the coordinates of the box where the defect is located.

[0104] Specifically, the image recognition module automatically learns and classifies the features in the image based on deep learning algorithms, especially convolutional neural networks (CNNs). By training the CNN model, the system can identify various defects in the image, such as cracks, deformations, color differences, etc. In addition, the image recognition module will also classify the defects in the image through the trained model and output the defect type and the specific position where the defect is located.

[0105] Convolutional neural network (CNN): Adopt a deep convolutional neural network to extract features in the image through multiple convolutional layers and pooling layers. Optimize the weights of the network through the backpropagation algorithm and the gradient descent method to achieve accurate classification of health product images.

[0106] Transfer learning: If there is a lack of sufficient labeled data, transfer learning can be adopted, starting from existing pre-trained models (such as ResNet, VGG), and performing fine-tuning to adapt to the characteristics of health product image data.

[0107] Beneficial effects:

[0108] Using deep learning technology can automatically identify and classify defects in the image, greatly reducing the cost and error of manual inspection.

[0109] By continuously training and updating the model, the system can continuously improve its recognition accuracy and adapt to the changing production environment and product characteristics.

[0110] In the defect detection and warning steps, defect evaluation comprehensively evaluates different types of defects according to the fuzzy logic control FLC or support vector SVM method, generates a quality report, and automatically triggers a warning signal according to a preset threshold;

[0111] In the defect detection and warning steps, the generated quality report includes not only defect information but also environmental data, production parameters, and adjustment history records of each production link for production management personnel to refer to;

[0112] The warning signals in the defect detection and warning steps include sound, visual, or email notifications.

[0113] Specifically, the main function of the defect detection and warning module is to conduct quality assessment based on the defect types provided by the image recognition module and generate warning signals according to the assessment results. If the severity of the detected defect exceeds the preset threshold, the system will automatically trigger production line adjustment or shutdown operations.

[0114] Defect assessment algorithm: Support Vector Machine (SVM) or Fuzzy Logic Control (FLC) is used for defect assessment. The impact of the defect on product quality is judged based on its type and size, and automatic decision-making is carried out in combination with production control rules.

[0115] Intelligent warning system: When a defect is identified and evaluated as unacceptable, the system will automatically issue a warning and notify production personnel through sound, visual signals or emails.

[0116] Beneficial effects:

[0117] Provide real-time quality monitoring and warning functions, which can detect and handle problems in a timely manner at the initial stage of the problem occurrence, and prevent unqualified products from flowing into the market.

[0118] Enhance the automation and intelligence of the production process, reduce human intervention, and improve production efficiency and quality consistency.

[0119] In the production control steps, the adjustment of production process parameters includes automatically adjusting the temperature, humidity and pressure parameters of production equipment, and precise control is carried out through the PID control algorithm or the adaptive control algorithm.

[0120] Specifically, the production control module automatically adjusts each link on the production line according to the output of the defect detection and warning module to ensure product quality. By controlling production equipment (such as granulators, packaging machines, etc.) to optimize production parameters such as temperature, humidity, production speed, etc.

[0121] PLC control system: Through the interface with the PLC (Programmable Logic Controller) system, the operation of the equipment is automatically controlled and the production process parameters are adjusted.

[0122] PID control: In the production process control, the PID (Proportion, Integral, Derivative) control algorithm is used to accurately adjust key parameters such as temperature, humidity, production speed, etc.

[0123] Beneficial effects:

[0124] Realize the automatic adjustment of the production process, reduce the need for manual intervention, and improve production efficiency.

[0125] Ensure the quality stability and consistency of each batch of products by adjusting production parameters in real time.

[0126] In the data storage and traceability step, the blockchain adopts the Ethereum or Hyperledger Fabric platform, which is responsible for the transparency, credibility, and immutability of the traceability data.

[0127] Specifically, the data storage and traceability module is responsible for storing production data, quality inspection results, defect information, etc. in a secure database, and ensuring the security and traceability of the data through blockchain technology. This module provides complete production information traceability for each health care product.

[0128] Blockchain technology: Store production data through the Hyperledger Fabric or Ethereum blockchain platform to ensure the immutability and transparency of the data. Production data for each batch of products can generate records in the blockchain and remain permanently traceable.

[0129] Database management system: Use database management systems such as MySQL, MongoDB, or Cassandra to store data, and provide data query and analysis functions through API interfaces.

[0130] Beneficial effects:

[0131] Provide comprehensive and transparent production process records and quality control data, enhancing consumers' trust in products.

[0132] Utilize blockchain technology to ensure the security and immutability of data, providing reliable traceability information for regulatory agencies and consumers.

[0133] Embodiment 2:

[0134] Refer to Figure 4 , in the second embodiment of the present invention, the present invention provides a health care product production monitoring and control system based on image recognition, including:

[0135] Data acquisition module, which real-time collects image data, temperature data, and environmental parameters of the health care product production line through industrial cameras, infrared imaging devices, and environmental sensors;

[0136] Image processing module, which performs denoising, enhancement, edge detection, and feature extraction on the collected image data to improve the accuracy of subsequent image analysis;

[0137] Image recognition module, which uses a convolutional neural network CNN to perform in-depth analysis on the processed image data, identify defects of health care products, and classify them;

[0138] Defect detection and warning module, which based on the identified defects, performs quality assessment and generates warning signals according to the severity of the defects, and automatically adjusts the production process;

[0139] The production control module automatically adjusts production process parameters according to the defect detection results, including temperature, humidity, and production speed.

[0140] The data storage and traceability module stores production data, defect information, and quality control parameters in a database and is responsible for the security and traceability of the data using blockchain technology.

[0141] Embodiment III

[0142] In the third embodiment of the present invention, based on the same inventive concept, a computer-readable storage medium is proposed. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the method for monitoring and controlling the production of health products based on image recognition in the above embodiment.

[0143] Embodiment IV

[0144] In the fourth embodiment of the present invention, based on the same inventive concept, a computer device is proposed. The terminal includes: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory and implement the method for monitoring and controlling the production of health products based on image recognition in the above embodiment.

[0145] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following well-known technologies in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0146] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A health product production monitoring and control method based on image recognition, characterized in that It includes the following steps: S1. Data acquisition: Real-time acquisition of image data, temperature data, and environmental parameters of the health product production line through industrial cameras, infrared imaging devices, and environmental sensors; S2. Image processing: Denoising, enhancing, and edge detection are performed on the acquired image data to extract the appearance features, shape, and size information of the health products; S3. Image recognition: Use a convolutional neural network CNN to analyze the processed images, identify defects in the health products, and classify the defects; S4. Defect detection and warning: Evaluate the product quality based on the identified defects, generate a quality report, and automatically issue a warning according to the severity of the defects, triggering automatic adjustment during the production process; S5. Production control: Automatically adjust the production process parameters according to the results of defect detection, including temperature, humidity, and production speed; S6. Data storage and traceability: Store the acquired production data, defect information, and quality control parameters in a database, and be responsible for the security and traceability of the data through blockchain technology.

2. The method for monitoring and controlling the production of health products based on image recognition according to claim 1, wherein, In the data acquisition step, the image data acquired by the industrial camera is a high-definition image, and the acquisition area of the industrial camera covers all links of the health product production line.

3. The method for monitoring and controlling the production of health products based on image recognition according to claim 1, wherein The image processing specifically includes: S201. Denoising processing: Apply median filtering or Gaussian filtering to the image. The median filtering window is 3x3 pixels or 5x5 pixels; Gaussian filtering uses a Gaussian kernel function with a standard deviation of 0.8 or 1.0 for smoothing processing; S202. Image enhancement: Use the adaptive histogram equalization algorithm to enhance the details of low-contrast image areas, increasing the contrast by 50% to 200%; S203. Edge detection: Use the Canny edge detection algorithm, set the minimum threshold to 100 and the maximum threshold to 200, or use the Sobel operator to extract the edge information in the image.

4. The method for monitoring and controlling the production of health products based on image recognition according to claim 1, characterized in that, The image recognition specifically includes the following steps: S301. Use a convolutional neural network CNN to extract low-level and high-level features in the image In the convolutional layer of the convolutional neural network, multiple convolutional kernels are applied to extract local features, including edges, corners, and textures; In the pooling layer, use max pooling or average pooling operations, and the pooling window is 2x2 or 3x3, which is used to reduce the size of the feature map and retain key information; The number of layers of the CNN network is 4 convolutional layers and 2 fully connected layers to extract image features of different scales and different levels; S302. Based on the labeled image dataset, optimize the weights of the convolutional neural network through the backpropagation algorithm and the gradient descent method The training dataset contains at least 5000 images, and each image contains a corresponding label, where each category contains at least 1000 images; Use batch gradient descent for training. Each batch contains 64 images. The Adam optimizer is used during training, and the learning rate is set to 0.001; Train the CNN model through 50 iterations, and evaluate the validation set during each iteration. Finally, the training accuracy exceeds 95%; S303. Perform defect detection and classification on the input image through the trained convolutional neural network Perform forward propagation on the input image. The CNN outputs the probability of each defect category. Use the Softmax function to normalize the output results and output the category probabilities. Select the category with the highest probability as the final defect type; The defect categories output by the CNN classifier include "crack", "deformation", "color difference", and "no defect". The probability value range for each category is from 0 to 1, and the sum of the probabilities is 1; Precisely locate the defects in the image through an object detection algorithm: Use the Faster R-CNN or YOLO algorithm to precisely locate the position of the defect by generating candidate boxes and further performing regression adjustment; The position of each candidate box is precisely adjusted through region regression, making the positioning error less than 10%; Output the position information of each defect and calibrate the coordinates of the box where the defect is located.

5. The method for monitoring and controlling the production of health products based on image recognition according to claim 1, wherein In the defect detection and warning step, defect evaluation comprehensively evaluates different types of defects according to the fuzzy logic control FLC or support vector SVM method, generates a quality report, and automatically triggers a warning signal according to a preset threshold; In the defect detection and warning step, the generated quality report includes not only defect information but also environmental data, production parameters, and adjustment history records of each production link for production management personnel to refer to; The warning signals in the defect detection and warning step include sound, visual, or email notifications.

6. The method for monitoring and controlling the production of health products based on image recognition according to claim 1, characterized in that, In the production control step, the adjustment of production process parameters includes automatically adjusting the temperature, humidity, and pressure parameters of the production equipment, and precise control is performed through the PID control algorithm or the adaptive control algorithm.

7. The method for monitoring and controlling the production of health products based on image recognition according to claim 1, wherein In the data storage and traceability step, the blockchain uses the Ethereum or Hyperledger Fabric platform, which is responsible for the transparency, credibility, and immutability of the traceability data.

8. A health product production monitoring and control system based on image recognition, characterized in that, For the method for monitoring and controlling the production of health products based on image recognition according to any one of claims 1-7, it includes: A data acquisition module that real-time collects image data, temperature data, and environmental parameters of the health product production line through industrial cameras, infrared imaging devices, and environmental sensors; An image processing module that performs denoising, enhancement, edge detection, and feature extraction on the collected image data to improve the accuracy of subsequent image analysis; An image recognition module that uses the convolutional neural network CNN to deeply analyze the processed image data, identify defects in health products, and classify them; A defect detection and warning module that, based on the identified defects, performs quality assessment and generates a warning signal according to the severity of the defects, and automatically adjusts the production process; A production control module that automatically adjusts production process parameters, including temperature, humidity, and production speed, according to the defect detection results; A data storage and traceability module that stores production data, defect information, and quality control parameters in a database, and uses blockchain technology to be responsible for the security and traceability of the data.

9. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for monitoring and controlling the production of health products based on image recognition according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that, A computer program is stored on the readable storage medium, and when the computer program is executed by a processor, it implements the method for monitoring and controlling the production of health products based on image recognition according to any one of claims 1 to 7.

Images