A method and device for supervising the batching of a quartz stone production line based on image recognition

Abstract

The present invention relates to the field of image data processing, and particularly to a method and device for supervising the batching of a quartz stone production line based on image recognition. The method and device include: installing a high-definition camera in the batching link of the quartz stone production line to collect images of the batching process in real time; preprocessing the images of the batching process; using a preset recognition model to extract key features from the preprocessed images of the batching process to determine the batching information; judging whether the batching information is abnormal according to the preset batching ratio information; when it is found that the batching is abnormal, automatically triggering an alarm mechanism to notify the operator in time for adjustment and processing. The technical problems of low efficiency and high error rate in the existing quartz stone production line batching supervision method are solved.

Description

Technical Field

[0001] The present invention relates to the field of image data processing, and in particular, to a method and device for supervising and managing the batching of a quartz stone production line based on image recognition. Background Art

[0002] As an important industrial raw material, quartz stone has a wide range of applications in the fields of building materials, chemical engineering, etc. The batching process of the quartz stone production line has a crucial impact on the quality and performance of the products. During the production process of quartz stone, there are often errors in the weighing of raw materials, which directly affect the product ratio and final quality. Inaccurate weighing may lead to unstable product performance and inability to meet customer requirements. At the same time, a large amount of dust is easily generated during the processing and batching of raw materials such as quartz stone, which may result in excessive impurities in the raw materials and affect the quality of quartz stone.

[0003] However, traditional batching supervision methods often rely on manual inspections and empirical judgments, and cannot detect and handle material leakage and misplacement in a timely manner, resulting in problems such as low efficiency and high error rates.

[0004] Therefore, how to provide a method and device for supervising and managing the batching of a quartz stone production line based on image recognition has important practical significance and application value. Summary of the Invention

[0005] The present invention provides a method and device for supervising and managing the batching of a quartz stone production line based on image recognition, which solves the problems of low efficiency and high error rates in the existing batching supervision methods for quartz stone production lines.

[0006] A method for supervising and managing the batching of a quartz stone production line based on image recognition according to the present invention specifically includes the following technical solutions:

[0007] Install a high-definition camera at the batching link of the quartz stone production line to collect images of the batching process in real time;

[0008] Preprocess the images of the batching process;

[0009] Use a preset recognition model to extract key features from the preprocessed images of the batching process and determine the batching delivery information;

[0010] Judge whether the batching delivery information is abnormal according to the preset batching ratio information;

[0011] When it is found that the delivered batching is abnormal, automatically trigger an alarm mechanism and notify the operator in time for adjustment and processing.

[0012] Further, it also includes: recording the obtained batching delivery information to provide a basis for subsequent quality control and quality traceability.

[0013] Further, the preprocessing of the images of the batching process includes: performing median filtering on the images of the batching process to obtain denoised images;

[0014] Among them, the expression of the median filtering is:

[0015] In the formula, is the denoised image output after median filtering, is the median filtering function, is the sliding window, is the two-dimensional data sequence of the image data, and i, j are coordinates.

[0016] Further, key features are extracted from the preprocessed images of the batching process by using a preset recognition model to determine the batching placement information, including:

[0017] Obtain standard quartz stone images;

[0018] Construct a recognition model, process the standard quartz stone image into a 3-channel image with a resolution of 224*224 and input it into the recognition model, and train and learn key features through the recognition model;

[0019] Input the preprocessed images of the batching process into the recognition model and output key features;

[0020] Determine the batching placement information according to the output key features.

[0021] Further, the recognition model includes convolutional neural sub-network models that extract different key features. Among them, multi-scale feature map fusion technology is used for feature fusion between each sub-network model.

[0022] Further, the structure of each sub-network model is composed of a convolutional layer, a pooling layer, a fully connected layer, and a SoftMax layer.

[0023] Further, when inputting the preprocessed images of the batching process into the recognition model to output key features, among them, the expression of the nth output key feature is:

[0024] In the formula, is the nth output key feature, is the nth convolution kernel, is the offset term, is the input image of the convolutional layer, represents two-dimensional convolution.

[0025] Further, the key features include: color feature, texture feature, and shape feature, where the texture feature includes quartz sand texture feature and resin texture feature.

[0026] Further, determining whether there is an abnormality in the ingredient feeding information according to the preset ingredient ratio information includes: using comparison software to compare the ingredient feeding information with the preset ingredient ratio information respectively. When the difference between the two is greater than the set threshold, it is determined that there is an abnormality in the ingredient feeding information; otherwise, there is no abnormality.

[0027] An apparatus for a method of supervising the ingredient feeding of a quartz stone production line based on image recognition includes: a display, a processor, and a computer program stored on a memory and executable on the processor. When the processor executes the program, the steps of the above method are implemented.

[0028] Positive and beneficial effects:

[0029] The multiple technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0030] 1. This method uses image recognition technology to monitor and identify the ingredient feeding process of the quartz stone production line in real time, which helps to improve the automation and intelligence level of the production line, improve production efficiency and product quality, reduce production costs, and enhance product quality and market competitiveness.

[0031] 2. By collecting image data during the ingredient feeding process and using advanced image recognition algorithms to analyze and process the images, key features and information are extracted, thereby realizing the automatic supervision of the ingredient feeding process. This method can not only improve the accuracy and efficiency of supervision, but also reduce the costs and risks of manual inspection, providing strong technical support for the ingredient management of the quartz stone production line.

[0032] 3. Through image recognition technology, real-time monitoring and precise management of the ingredient feeding process are achieved, solving the defects in the traditional quartz stone production process, improving production efficiency and product quality, and promoting the sustainable development of the quartz stone production industry.

[0033] 4. By using a convolutional neural sub-network model to automatically extract features from the images of the ingredient feeding process and gradually fuse them, it is beneficial to the abstract features for classification, breaking through the limitations of the traditional manual feature extraction method. Description of the Drawings

[0034] Figure 1 It is a flowchart of a method for supervising the ingredient feeding of a quartz stone production line based on image recognition provided by an embodiment of the present invention. Detailed Embodiments

[0035] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. 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 belong to the scope of protection of the present invention.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0037] The following specifically describes the specific solutions of a method and device for supervising the batching of a quartz stone production line based on image recognition provided by the present invention in conjunction with the accompanying drawings.

[0038] Referring to Figure 1 , which shows a method for supervising the batching of a quartz stone production line based on image recognition provided by an embodiment of the present invention, specifically including the following steps:

[0039] Install a high-definition camera in the batching link of the quartz stone production line to collect images of the batching process in real time;

[0040] Preprocess the images of the batching process;

[0041] Use a preset recognition model to extract key features from the preprocessed images of the batching process and determine the batching placement information;

[0042] According to the preset batching ratio information, determine whether the batching placement information is abnormal;

[0043] When it is found that the placed batching is abnormal, automatically trigger an alarm mechanism and notify the operator in time for adjustment and handling.

[0044] Specifically, it further includes: recording the obtained batching placement information to provide a basis for subsequent quality control and quality traceability.

[0045] Specifically, preprocessing the images of the batching process includes: performing median filtering on the images of the batching process to obtain denoised images;

[0046] Among them, the expression of median filtering is:

[0047] In the formula, is the denoised image output after median filtering, is the median filtering function, is the sliding window, It is a two-dimensional data sequence of image data, where i and j are coordinates.

[0048] Median filtering can effectively remove various types of noise such as salt-and-pepper noise and Gaussian noise in images. Since median filtering does not depend on the absolute magnitude of pixel values but takes the median of pixel values within the window, it is relatively less affected by outliers (such as noise points). At the same time, compared with other linear filtering methods (such as mean filtering), median filtering can better preserve the edge information of the image while denoising. Also, median filtering can be used to reduce the sharpening and enhancement effects in images.

[0049] Specifically, use a preset recognition model to extract key features from the preprocessed images of the batching process, and determine the batching information, including:

[0050] Obtain a standard quartz stone image;

[0051] Construct a recognition model, process the standard quartz stone image into a 3-channel image with a resolution of 224*224 and input it into the recognition model, and train and learn key features through the recognition model;

[0052] Input the preprocessed images of the batching process into the recognition model and output key features;

[0053] Determine the batching information according to the output key features.

[0054] Specifically, the recognition model includes convolutional neural sub-network models that extract different key features. Among them, multi-size feature map fusion technology is used for feature fusion between each sub-network model.

[0055] Specifically, the structure of each sub-network model is composed of a convolutional layer, a pooling layer, a fully connected layer, and a SoftMax layer.

[0056] Specifically, the convolutional neural network is one of the most commonly used models in the field of deep learning for image recognition. It automatically learns and extracts hierarchical features in images through a series of convolutional layers, pooling layers, and fully connected layers. During the training process, the convolutional neural network can learn from low-level edge and texture features to high-level object and scene features, thereby achieving a deep understanding of the image content.

[0057] Specifically, input the preprocessed images of the batching process into the recognition model and output key features. Among them, the expression of the nth output key feature is:

[0058] In the formula, is the nth output key feature, is the nth convolutional kernel, is the offset term, is the input image of the convolutional layer, is represented as a two-dimensional convolution.

[0059] Specifically, the key features include: color features, texture features, and shape features. Among them, the texture features include quartz sand texture features and resin texture features.

[0060] Specifically, the texture features of quartz sand: Quartz sand is one of the main components in quartz stone, and its texture features mainly depend on the particle shape, color, and size of quartz sand. Quartz sand usually presents a granular texture, with obvious particle boundaries and distribution rules between particles. The color of quartz sand can vary according to different mineral compositions, and common colors include white, gray, pink, etc. Quartz sand has a high hardness, is hard in texture, and usually has strong wear resistance and corrosion resistance.

[0061] The texture features of resin: Resin is the binder in quartz stone, used to bond quartz sand and other raw materials such as pigments together. The texture features of resin are mainly manifested in its smooth and uniform surface texture, usually presenting a uniform and smooth coating-like structure. The color of resin is usually transparent or semi-transparent, and pigments can be added according to needs to adjust the color.

[0062] Specifically, the multi-size feature map fusion technology is used for feature fusion between each sub-network model, including:

[0063] Each sub-network model will extract features from its input data. After feature extraction, each sub-network model will generate one or more feature maps. These feature maps may have different spatial resolutions, that is, different sizes. Smaller feature maps usually contain higher-level semantic information, while larger feature maps retain more details and spatial information. Next, the multi-size feature maps from different sub-network models are fused. This can be achieved in various ways, for example: direct concatenation: Concatenate feature maps of different sizes in the channel dimension to form a feature map with a larger number of channels. Upsampling / downsampling: Upsample the smaller feature map or downsample the larger feature map to make them have the same spatial resolution, and then fuse them. Attention mechanism: Use the attention mechanism to learn the weights between different feature maps to achieve dynamic feature fusion.

[0064] Specifically, through feature fusion, the key features of each type of quartz stone are used as observation vectors, and the Viterbi algorithm is used to find the output probability corresponding to each sub-network model, and then they are concatenated into a one-dimensional vector, which is the output probability of the nth key feature in the mth class in the training set under the nth sub-network model.

[0065] Specifically, according to the preset ingredient ratio information, determine whether there is an abnormality in the ingredient feeding information, including: using comparison software to compare the ingredient feeding information with the preset ingredient ratio information respectively. When the difference between the two is greater than the set threshold (the specific threshold is selected according to the actual situation), it is determined that there is an abnormality in the ingredient feeding information, such as material leakage, wrong material, etc., and it is discovered and processed in time, otherwise there is no abnormality.

[0066] An apparatus for a quartz stone production line ingredient supervision method based on image recognition, including: a display, a processor, and a computer program stored on a memory and executable on the processor. When the processor executes the program, the steps of the method are implemented.

[0067] In summary, a quartz stone production line ingredient supervision method and apparatus based on image recognition are completed, which helps to improve the automation and intelligence level of the quartz stone production line, and improve production efficiency and product quality.

[0068] The order of the invention embodiments is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0069] Each embodiment in this specification is described in a progressive manner. The same or similar parts between each embodiment can be referred to each other. The key point of each embodiment is to illustrate the differences from other embodiments.

[0070] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for supervising the batching of a quartz stone production line based on image recognition, characterized in that, It includes the following steps: Install a high-definition camera in the batching link of the quartz stone production line to collect images of the batching process in real time; Preprocess the images of the batching process; Extract key features from the preprocessed images of the batching process using a preset recognition model, and determine the batching information, including: obtaining standard quartz stone images; constructing a recognition model, processing the standard quartz stone images and inputting them into the recognition model, and training and learning key features through the recognition model; inputting the preprocessed images of the batching process into the recognition model to output key features; determining the batching information according to the output key features; wherein, the recognition model includes convolutional neural sub-network models for extracting different key features, and multi-scale feature map fusion technology is used for feature fusion between each sub-network model; through feature fusion, the key features of each type of quartz stone are used as observation vectors, and the output probability corresponding to each sub-network model is obtained using the Viterbi algorithm, and then they are concatenated into a one-dimensional vector, the output probability of the nth key feature in the mth class in the training set under the nth sub-network model; the output nth key feature: ; is the output nth key feature, is the nth convolutional kernel, is the offset term, is the input image of the convolutional layer, is the two-dimensional convolution; Judge whether there is an abnormality in the batching input information according to the preset batching ratio information; When it is found that the input batching is abnormal, automatically trigger the alarm mechanism and notify the operator in time for adjustment and processing.

2. The batching supervision method for a quartz stone production line based on image recognition according to claim 1, wherein, It also includes: Record the obtained batching input information.

3. A method for supervising the batching of a quartz stone production line based on image recognition according to claim 1, characterized in that, The preprocessing of the images of the batching process includes: performing median filtering on the images of the batching process to obtain denoised images; Among them, the expression of the median filtering is: ； In the formula, is the denoised image output after median filtering, is the median filtering function, is the sliding window, is the two-dimensional data sequence of the image data, and i, j are coordinates.

4. The batch monitoring method for a quartz stone production line based on image recognition according to claim 1, wherein, The structure of each sub-network model is composed of a convolutional layer, a pooling layer, a fully connected layer and a SoftMax layer.

5. The batching supervision method for a quartz stone production line based on image recognition according to claim 1, characterized in that, The key features include: color features, texture features, shape features, where the texture features include quartz sand texture features and resin texture features.

6. The batching supervision method for a quartz stone production line based on image recognition according to claim 1, wherein The judgment of whether there is an abnormality in the batching input information according to the preset batching ratio information includes: using comparison software to compare the batching input information with the preset batching ratio information respectively. When the difference between the two is greater than the set threshold, it is judged that there is an abnormality in the batching input information, otherwise there is no abnormality.

7. An apparatus for a batching supervision method of a quartz stone production line based on image recognition, comprising: A display, a processor, and a computer program stored on a memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 6 when executing the program.

Images