Machine vision-based separator screen surface separation detection method and system

Abstract

The application relates to the technical field of machine vision, in particular to a separation machine screen surface separation detection method and system based on machine vision. The method comprises the following steps: deploying an image acquisition component based on a screen surface light illumination adaptability model, performing texture enhancement processing on screen surface images collected by the image acquisition component to obtain a texture enhancement factor; extracting a particle distribution density according to the texture enhancement factor, and constructing a global separation consistency index of the entire screen surface state according to the particle distribution density; generating a saliency weight according to the global separation consistency index; and generating a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold. The application strengthens the screen surface texture features in a complex particle motion background by introducing a multi-scale texture enhancement and dynamic block mechanism, finely identifies the abnormal areas of the screen surface, intelligently classifies the defect types, and improves the accuracy and stability of the abnormal detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of machine vision, in particular to a separation machine screen separation detection method and system based on machine vision. BACKGROUND

[0002] The separation machine is a device for separating different components in a mixture by physical or mechanical methods, and is a key device in the material screening process, including commonly used industrial vibrating screen separation machines, centrifugal separation machines or air flow classification separation machines. The multi-layer vibrating screen separation machine used in the coal preparation plant, or the high-speed centrifugal screen used for classification of fine particle materials in the ore processing process.

[0003] The running state of the screen of the separation machine directly affects the separation efficiency and product quality. The existing separation detection method generally relies on vibration sensors or fixed cameras for image acquisition, and uses traditional image processing algorithms or simple threshold methods to identify screen abnormalities such as blockage, unbalanced load or screen hole blockage. However, such methods generally have insufficient detection accuracy, poor anti-interference ability, and lack of dynamic trend analysis. Especially in the case of high dynamic, high occlusion and high particle motion interference on the screen, the existing technology often cannot accurately identify local abnormalities, and it is also difficult to make timely warning on the evolution trend of the abnormality. In addition, the distribution of material particles on the screen is complex, and the existing technology has limited use of explicit features such as particle morphology and distribution stability. There is a lack of targeted feature modeling and multi-source feature fusion mechanism, resulting in weak abnormal type discrimination ability, high misjudgment rate, and difficulty in meeting the needs of fine perception and prediction control of the screen running state under the condition of intelligent manufacturing. SUMMARY

[0004] Therefore, it is necessary to provide a separation machine screen separation detection method and system based on machine vision, which introduces a multi-scale texture enhancement and dynamic block mechanism to strengthen the texture features of the screen under complex particle motion background, effectively improve the particle boundary recognition ability, and accurately identify the screen separation abnormal area and intelligently classify the defect type, improve the accuracy and stability of the abnormal detection, effectively distinguish different screen abnormal types such as blockage and unbalanced load, and solve the problem of lack of detailed classification of abnormalities and high misjudgment rate in the prior art.

[0005] The technical scheme of the present application is as follows:

[0006] A separation machine screen separation detection method based on machine vision, the method comprising:

[0007] Deploying an image acquisition component based on a screen light adaptation model, and performing texture enhancement processing on the screen image collected by the image acquisition component to obtain a texture enhancement factor;

[0008] extract a particle distribution density according to the texture enhancement factor, and construct a global separation consistency index of the whole screen surface state according to the particle distribution density;

[0009] generate a saliency weight according to the global separation consistency index;

[0010] generate a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold.

[0011] Specifically, an image acquisition component is deployed based on a screen surface light adaptation model, and a screen surface image collected by the image acquisition component is subjected to texture enhancement processing to obtain a texture enhancement factor, including:

[0012] A screen surface light adaptation model under light reflection conditions is constructed, and the light source angle and intensity of the image acquisition component are adjusted according to the model to deploy the image acquisition component;

[0013] The screen surface image collected by the image acquisition component is subjected to dynamic block and multi-scale texture enhancement processing to obtain the texture enhancement factor.

[0014] Specifically, a particle distribution density is extracted according to the texture enhancement factor, and a global separation consistency index of the whole screen surface state is constructed according to the particle distribution density, including:

[0015] A dynamic texture weight model is set according to the texture enhancement factor, a quantifiable image index for a separation state is constructed, and a particle distribution density is extracted;

[0016] A global separation consistency index of the whole screen surface state is constructed based on the extracted particle distribution density.

[0017] Specifically, a saliency weight is generated according to the global separation consistency index, including:

[0018] A local abnormal saliency evaluation model is constructed according to the global separation consistency index;

[0019] The saliency weight is output based on the local abnormal saliency evaluation model.

[0020] Specifically, the method further includes:

[0021] A region with a serious separation anomaly in the screen surface image is identified based on the saliency weight.

[0022] Specifically, the screen surface defect judgment threshold includes a first abnormal saliency threshold, a second abnormal saliency threshold, and a morphological stability judgment threshold;

[0023] A screen surface defect judgment result is generated according to the saliency weight and a preset screen surface defect judgment threshold, including:

[0024] construct a classification function according to the saliency weight and the first abnormal saliency threshold, the second abnormal saliency threshold and the morphological stability determination threshold;

[0025] generate a screen surface defect determination result according to the classification function.

[0026] Specifically, a machine vision-based screen surface separation detection system, the system comprising:

[0027] a texture enhancement factor generation module configured to deploy an image acquisition component based on a screen surface light adaptation model, and perform texture enhancement processing on screen surface images collected by the image acquisition component to obtain a texture enhancement factor;

[0028] a consistency index generation module configured to extract a particle distribution density according to the texture enhancement factor, and construct a global separation consistency index of the entire screen surface state according to the particle distribution density;

[0029] a saliency weight generation module configured to generate a saliency weight according to the global separation consistency index;

[0030] a defect determination result generation module configured to generate a screen surface defect determination result according to the saliency weight and a preset screen surface defect determination threshold.

[0031] Specifically, the texture enhancement factor generation module is further configured to:

[0032] construct a screen surface light adaptation model under light reflection conditions, adjust the light source angle and intensity of the image acquisition component according to the model to deploy the image acquisition component, and perform dynamic block and multi-scale texture enhancement processing on the screen surface images collected by the image acquisition component to obtain the texture enhancement factor.

[0033] Specifically, the consistency index generation module is further configured to: set a dynamic texture weight model according to the texture enhancement factor, construct a quantifiable image index for the separation state, and extract the particle distribution density; and construct the global separation consistency index of the entire screen surface state based on the extracted particle distribution density.

[0034] Specifically, the saliency weight generation module is further configured to: construct a local abnormal saliency evaluation model according to the global separation consistency index; and output a saliency weight based on the local abnormal saliency evaluation model.

[0035] Specifically, the saliency weight generation module is further configured to: identify a region with severe separation abnormality in the screen surface image based on the saliency weight.

[0036] Specifically, the screen surface defect judgment threshold includes a first abnormal saliency threshold, a second abnormal saliency threshold and a shape stability judgment threshold; the defect judgment result generation module is further configured to: construct a classification function according to the saliency weight and the first abnormal saliency threshold, the second abnormal saliency threshold and the shape stability judgment threshold; and generate a screen surface defect judgment result according to the classification function.

[0037] Optionally, a computer device is also provided, including a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above-mentioned machine vision-based screen surface separation detection method when executing the computer program.

[0038] Optionally, a computer readable storage medium is also provided, which stores a computer program, and the computer program implements the steps of the above-mentioned machine vision-based screen surface separation detection method when executed by a processor.

[0039] The present application relates to machine learning and deep learning technology, which realizes the following technical effects:

[0040] (1) The machine vision-based screen surface separation detection method and system of the present application deploy an image acquisition component based on a screen surface light adaptation model, perform texture enhancement processing on the screen surface images collected by the image acquisition component, obtain a texture enhancement factor, adjust the image acquisition conditions by constructing a light adaptation model, ensure the light balance and detail clarity of the screen surface images, and significantly improve the image quality and the robustness of subsequent analysis.

[0041] (2) The particle distribution density is extracted according to the texture enhancement factor, a global separation consistency index of the entire screen surface state is constructed according to the particle distribution density, a saliency weight is generated according to the global separation consistency index, and a screen surface defect judgment result is generated according to the saliency weight and a preset screen surface defect judgment threshold. The introduction of the multi-scale texture enhancement and dynamic block mechanism can strengthen the screen surface texture features in a complex particle motion background, effectively improve the particle boundary recognition ability, construct a screen surface separation consistency index based on the extracted particle distribution density, break the traditional isolated judgment mode of local abnormalities, and realize the quantitative description of the global state of the screen surface.

[0042] (3) The present application also combines a local abnormal saliency evaluation model and a classification function to finely identify and intelligently classify the defect types of the screen surface separation abnormal area, not only improves the accuracy and stability of abnormal detection, but also effectively distinguishes different screen surface abnormal types such as blockage and unbalanced load, and solves the problem of lack of detailed classification and high misjudgment rate of existing technologies. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 Fig. 2 is a flowchart illustrating a method for detecting screen separation of a separator according to an embodiment of the present disclosure;

[0044] Figure 2 Fig. 3 is a block diagram illustrating a system for detecting screen separation of a separator according to an embodiment of the present disclosure;

[0045] Figure 3 Fig. 4 is a block diagram illustrating a computer device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0046] In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular sequences of steps, techniques, etc., in order to provide a thorough understanding of the embodiments of the present disclosure. However, it will be apparent to those skilled in the art that the present embodiments can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known methods, devices, and circuits are omitted so as not to obscure the description of the present embodiments.

[0047] It is to be understood that the terminology "includes", "has", "holds", "contains" and / or "comprising", when used in this specification and in the following claims, indicates the presence of the described features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0048] It is also to be understood that the terminology "and / or" when used in this specification and in the following claims, refers to at least one of the items, or any combination of the items, and includes all possible combinations when dependent on two or more items.

[0049] As used in this specification and in the claims, the term "if" can be interpreted as meaning "when", or "once", or "in response to a determination", or "in response to detecting", as appropriate, depending on the context. Similarly, the phrase "if determined", or "if detected [the described condition or event]" can be interpreted as meaning "once determined", or "in response to a determination", or "once detected [the described condition or event]", or "in response to detecting [the described condition or event]", as appropriate, depending on the context.

[0050] In addition, in the description of the specification and the appended claims, the terms "first", "second", "third", etc. are used only to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

[0051] Reference throughout this application to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase "in one embodiment" or "in some embodiments" in various places throughout this specification are not necessarily all referring to the same embodiment, however, are meant to signify that a particular feature, structure, or characteristic described is included in at least one embodiment. The terms "including," "comprising," "having," and variations thereof are meant to encompass the items listed thereafter, but do not exclude additional items from also being present. The terms "a" and "an" are meant to encompass both the singular and the plural.

[0052] In one embodiment, a terminal is provided, configured to: deploy an image acquisition component based on a screen surface lightness adaptability model, and perform texture enhancement processing on screen surface images acquired by the image acquisition component to obtain a texture enhancement factor; extract a particle distribution density according to the texture enhancement factor, and construct a global separation consistency index of the entire screen surface state according to the particle distribution density; generate a saliency weight according to the global separation consistency index; and generate a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold.

[0053] The terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.

[0054] In one embodiment, as shown in Figure 1 A machine vision-based screen surface separation detection method is provided, including:

[0055] Step S100: deploy an image acquisition component based on a screen surface lightness adaptability model, and perform texture enhancement processing on screen surface images acquired by the image acquisition component to obtain a texture enhancement factor;

[0056] Step S200: extract a particle distribution density according to the texture enhancement factor, and construct a global separation consistency index of the entire screen surface state according to the particle distribution density;

[0057] Step S300: generate a saliency weight according to the global separation consistency index;

[0058] Step S400: generate a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold.

[0059] The embodiment is based on a screen surface illumination adaptability model to deploy an image acquisition component, and performs texture enhancement processing on a screen surface image acquired by the image acquisition component to obtain a texture enhancement factor; extracts a particle distribution density according to the texture enhancement factor, and constructs a global separation consistency index of the entire screen surface state according to the particle distribution density; generates a saliency weight according to the global separation consistency index; and generates a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold. By introducing a multi-scale texture enhancement and dynamic block mechanism, the application strengthens the screen surface texture features in a complex particle motion background, performs fine identification on the screen surface separation abnormal area, and intelligently classifies the defect types, thereby improving the accuracy and stability of the abnormal detection.

[0060] In one embodiment, step S100: based on a screen surface illumination adaptability model, deploying an image acquisition component, and performing texture enhancement processing on a screen surface image acquired by the image acquisition component to obtain a texture enhancement factor, comprises:

[0061] Step S110: constructing a screen surface illumination adaptability model under illumination reflection conditions, adjusting the light source angle and intensity of the image acquisition component according to the model, to deploy the image acquisition component.

[0062] Step S120: performing dynamic block and multi-scale texture enhancement processing on the screen surface image acquired by the image acquisition component to obtain a texture enhancement factor.

[0063] In the embodiment, in order to provide input image data for subsequent screen surface feature extraction, the core target cannot only be to acquire images, but also needs to combine the reflection characteristics of the screen surface material and the material particles to realize parameterized modeling of the illumination reflection conditions. Therefore, a screen surface illumination adaptability model under illumination reflection conditions needs to be constructed, and the light source angle and intensity of the image acquisition component need to be adjusted according to the model to deploy the image acquisition component.

[0064] The screen surface illumination adaptability model uses the average brightness, particle boundary definition, and noise background ratio of the screen surface area to construct a screen surface illumination adaptability index. The screen surface illumination adaptability model is as follows:

[0065]

[0066] Wherein, is the screen surface illumination adaptability. is the average gray value of the unit area screen surface image, which is calculated by statistically averaging the pixel gray values in the unit area, and then calculating after converting the gray image or the color image to a gray image. The average gray value of the unit area screen surface image is used to reflect the overall brightness level of the area, and measure the uniformity of the illumination; is the image edge definition coefficient, the edge of the image is detected by Sobel or Canny operator, and the average intensity of the edge response value in the unit area is counted. Reflect whether the boundary of the particle or the screen hole in the area is clear, which affects the image separability. is the background noise ratio, the signal-to-noise ratio is calculated by comparing the standard deviation of the background area with the gray value of the main area of the image. It indicates the interference degree of non-target area in the image to the recognition effect, and the more complex the background is, the higher the value is. is a small constant set to prevent the denominator from being zero, which is set to a fixed constant much smaller than 1 by those skilled in the art, such as 0.001, which is only used to ensure the numerical stability of the model calculation, and does not depend on the image content.

[0067] The image acquisition component will adjust the light source angle and intensity according to the criterion of maximizing , select the image state most conducive to subsequent image segmentation and feature extraction, and ensure that the quality of the subsequent input image reaches the optimal state. The prior art often ignores the reflection difference between the screen material such as metal and rubber and the material particles such as coal powder and gravel under different light angles, resulting in blurred texture details or difficult edge recognition in the image, especially in scenes with strong reflection or high noise background. In contrast, the screen surface illumination adaptability model in the embodiment automatically performs a round of illumination scanning after the image acquisition component is deployed, that is, adjusts different light source angles and intensities, collects corresponding images, and calculates the illumination adaptability index of each image in real time ; then, according to the size of , it is judged whether the current configuration is optimal, if the value of does not reach the preset threshold, the parameters are adjusted to continue to collect, until the optimal or convergent stable is reached, and the current illumination condition is locked for formal sampling. The preset threshold corresponding to can be set by experimental calibration, that is, a plurality of groups of images are collected under typical screen surface and typical particle conditions, and the lower limit value of the illumination adaptability that can meet the clear edge recognition and low noise interference is calculated. In practical application, an empirical value between 15.0 and 20.0 can be set, and the specific value is set by those skilled in the art, and the present application does not make specific limitations.

[0068] The method disclosed in the embodiment can dynamically adjust parameters according to real-time illumination changes in complex dust environments such as coal mines, sandstone plants, and metallurgical workshops, to ensure the stability of image acquisition. The system can optimize image quality by adjusting the directionality of the light source array, the light supplement ratio, and the exposure parameters of the acquisition equipment under extreme reflection conditions such as metal screen plates or low illumination conditions such as night work.

[0069] Further, in order to solve the problem that the image local area information is easily lost due to particle accumulation, shielding, local reflection and the like in a complex screen surface environment due to the use of a full-image uniform enhancement strategy in the prior art, and the noise amplification or edge blur region cannot be effectively improved in image separability due to uniform enhancement, in step S120, the screen surface image collected by the image collection component is subjected to dynamic blocking and multi-scale texture enhancement processing to obtain a texture enhancement factor.

[0070] Specifically, the dynamic blocking processing is to dynamically divide the whole task into smaller "blocks" for independent processing according to the characteristics of the data or the resource status at runtime, so as to improve efficiency, adapt to changes or enhance robustness. In the embodiment, the brightness change rate, particle distribution gradient and gray contrast are weighted and judged to accurately segment different regions.

[0071] In each image blocking region , the corresponding texture enhancement factor is as follows:

[0072]

[0073] wherein, is the screen surface light adaptation degree output by step S110, and is a stability weight at the full-image level.

[0074] is the average value of the gray gradient amplitude of the region , which reflects the local texture change intensity. The Sobel operator is applied to the image region , the average value of the gradient amplitude of each pixel is calculated, and is used to represent the texture change intensity. is the skewness of the gray distribution of the region, which measures whether the texture distribution of the region is balanced. The skewness (Skewness) of all pixel gray values in the region is calculated to reflect the symmetry of the gray distribution, and the third central distance formula is commonly used. is the texture enhancement factor, which represents whether the region should be enhanced in texture extraction, and is the decision basis for whether the region should be enhanced.

[0075] By calculating the of each region, if the value is greater than the texture response threshold set by the system, the region is determined to be a "weak texture area" and needs to be enhanced. The enhancement operations include local contrast stretching, edge response weighting, and directional filter (such as Gabor) application. All enhancement operations are performed on the The control is carried out to prevent false boundary generation caused by over-enhancement. The texture response threshold can be set according to the actual device resolution. Alternatively, the setting of the texture response threshold can be based on statistical analysis of sample images, such as calculating the value distribution of a large number of screen images under different states, and selecting the minimum value of the region with more than 80% clear texture as the threshold, so as to ensure that only the weak texture region is enhanced. In practical applications, if the device resolution is 1080p and the particle size is medium, the threshold can be set to 1.2-1.5, for example, when it is set to 1.3, the system only enhances the region less than the value, ensuring that the enhancement resources are concentrated on the low-quality image block.

[0076] In addition, using the model can also enable the system to make intelligent decisions on "enhancement budget allocation" according to the image state, that is, the limited computing resources are concentrated on the region with high value, thereby improving the overall computing efficiency and image quality.

[0077] In summary, the present step further expands the dimension of screen image processing, so that it not only focuses on the global quality of the image, but also can perform fine-grained structure analysis on local complexity. For example, when there are partial clumps, accumulation or wet material shielding on the screen, ordinary enhancement processing often cannot effectively restore the details, while the method in the present step S120 can identify these low-texture activity regions and weighted enhancement, so that the subsequent feature extraction and defect recognition are more accurate.

[0078] In one embodiment, step S200: extracting a particle distribution density according to the texture enhancement factor, and constructing a global separation consistency index of the entire screen state according to the particle distribution density, including:

[0079] Step S210: setting a dynamic texture weight model according to the texture enhancement factor, constructing a quantifiable image index for the separation state, and extracting a particle distribution density;

[0080] Step S220: constructing a global separation consistency index of the entire screen state based on the extracted particle distribution density.

[0081] In the present embodiment, in order to solve the problem that most of the existing screen state recognition methods only perform rough statistics on the particle distribution of the screen through simple pixel binary or edge recognition-based methods, and cannot fully reflect the complex working conditions such as local accumulation, hole blocking or unbalanced load. In the present embodiment, a dynamic texture weight model is set according to the texture enhancement factor, a quantifiable image index for the separation state is constructed, and a particle distribution density is extracted; then, based on the extracted particle distribution density, a global separation consistency index of the entire screen state is constructed.

[0082] The dynamic texture weight model is as follows: ​

[0083]

[0084] wherein, is a texture enhancement factor, reflecting the image quality and information richness of the region. is the number of particles identified in the region, which can be output by the machine vision detection module, i.e., the image acquisition component. is the area of the region, representing the area range per unit area, i.e., the area range corresponding to one image block region. is the actual particle distribution density in the region, which is also used to indirectly estimate the screen hole void ratio of the region. Specifically, by setting a mapping relationship between the unit particle coverage area and the weighted particle density , the screen hole occlusion ratio can be converted, thereby indirectly estimating the screen hole void ratio. By using to estimate the available void ratio of the screen hole region, it can be determined whether there is a particle accumulation, hole blocking, or material unloading screening abnormality in the region. After image blocking, the number of particles in each region is counted , and combined with the area of the region and the texture enhancement factor from step S2, the value of is calculated. The larger the value, the more densely the particles cover the region, and the image quality is good enough to have high analysis reliability. Then, a screen hole visual saturation threshold is set, and if exceeds this threshold, it is preliminarily determined as a "hole blocking or overloading" region, otherwise it is a normal screening region. This method realizes density-weighted estimation based on image readability, which has higher stability and accuracy than traditional methods based only on quantity statistics. The screen hole visual saturation threshold can be obtained by statistics on the normal screening state , and the upper limit of 90%-95% is taken as the critical value, for example, set to 0.045 particles / pixel 2 , which is determined as a local hole blocking or overloading region if it exceeds the value.

[0085] Therefore, by introducing the texture enhancement factor into the particle density modeling, this embodiment not only considers the number of particles, but also takes the image quality and readability of the region as a weighted index to construct a dynamic texture weight model. In addition, traditional models often ignore the visual availability of the screen hole, while this method further estimates the void degree of the screen hole in the region based on particle identification, i.e., the degree of screen surface coverage by particles. This void estimation can provide a quantitative basis for subsequent screening efficiency evaluation and abnormality judgment, and has high practical value.

[0086] Then, in order to solve the problem that the existing screen surface detection method is mostly limited to local area particle accumulation identification or simple average value statistics, and cannot systematically evaluate whether the whole screen surface running state has global separation unevenness, material flow deflection or screen hole partition blockage, etc., the regional particle density index output in the previous step is fused and analyzed in the whole screen surface range, and a global consistency index quantifies the stability and balance of the material separation on the screen surface. By statistically analyzing the deviation of each regional density value from the overall average density, not only can the blockage be detected, but also the early imbalance trend of the screen surface can be identified, realizing more proactive fault warning and separation quality control. Compared with the traditional method of detecting only the total amount of particles or the local maximum value, the model has higher robustness and stronger application universality.

[0087] The global separation consistency index is used to determine whether there is separation unevenness, local blockage or non-uniform arrangement problem. Specifically, the global separation consistency index is as follows:

[0088]

[0089] wherein, is the particle distribution density. is the average value of all , that is, the particle distribution reference of the whole screen surface. is the number of regions divided by the image, representing the number of regions divided by the image in the vertical direction (row), N represents the number of regions divided by the image in the horizontal direction (column), and M·N is the total number of regions divided by the image. is the variance weighted consistency evaluation index of the particle distribution density, that is, the global separation consistency index, which has the same dimension as . Through the global separation consistency index, the uniformity of the material distribution state in the running process of the separator screen surface is indirectly reflected. If is significantly greater than the threshold value, it indicates that there is a risk of regional particle accumulation or local blockage, and then the next stage of abnormal significance modeling and screen separation defect classification is entered. Wherein, the corresponding threshold value can be calculated by calculating the value distribution of a plurality of normal screen surface sample images, and taking the average value plus one standard deviation as the separation unevenness warning threshold value, which can be set to 0.006 particles / pixel 2 , and if the value exceeds the value, it is determined that the screen surface has obvious particle accumulation or unevenness.

[0090] In this embodiment, non-ideal states in industrial operation are particularly considered, for example, due to material adhesion, screen hole wear or uneven feeding, the particle density of part of the screen surface may deviate seriously. In this scenario, by The index can effectively detect whether the abnormal area has a systematic deviation or a developing accumulation trend, realizing the leap from local analysis to global judgment.

[0091] In one embodiment, step S300: generating a saliency weight according to the global separation consistency index, comprises:

[0092] Step S310: constructing a local anomaly saliency evaluation model according to the global separation consistency index;

[0093] Step S320: outputting a saliency weight based on the local anomaly saliency evaluation model.

[0094] Step S330: identifying a region with a serious separation anomaly in the screen surface image based on the saliency weight.

[0095] In this embodiment, in order to solve the problem that the traditional screen surface anomaly detection is mainly based on single particle density or texture features, it is difficult to consider the separation deviation and image complexity of the abnormal area at the same time, resulting in a high false positive rate, and the anomaly recognition ability under complex screen surface working conditions is insufficient, therefore, in this embodiment, a local anomaly saliency evaluation model is constructed according to the global separation consistency index, and a saliency weight is output based on the local anomaly saliency evaluation model.

[0096] Specifically, the local anomaly saliency evaluation model is as follows:

[0097]

[0098] wherein, is the particle distribution density. is the average value of . is the global separation consistency index. is a very small number to prevent the denominator from being zero. is a region texture complexity index, which is calculated by local direction gradient change, specifically, it is obtained by statistics of image gradient direction in the region, such as Sobel direction angle, or variance of local structure tensor, the larger the value is, the more complex the texture is, the more uneven the structure is, and the richer the image detail is and the more diverse the texture is. is a saliency weight, which quantifies the confidence and risk level of the anomaly of the region. Using classifies each region of the screen surface and extracts the abnormal region with serious separation. First, the relative deviation of the particle density of each region and the overall average density is calculated, and the standardized abnormal deviation coefficient is obtained by dividing the overall consistency index smoothing result. Then, the deviation coefficient is multiplied by the texture complexity index , which comprehensively considers the particle distribution anomaly and image texture features, thereby improving the reliability of anomaly recognition. According to The size of the abnormal region is ranked according to the risk level, and the high-risk abnormal region is screened out.

[0099] Specifically, first, the numerical distribution of all regions is counted The risk level is divided based on the quantile threshold of the distribution, such as 85%, 95%, for example, regions higher than the 95% quantile value are defined as "high risk", 85%-95% are "medium risk", and lower than 85% are "low risk". At the same time, it can be combined with the position of the region in the screen surface, such as concentrated in the discharge end or the central region, and further weighted optimization, so as to accurately screen out the high-risk abnormal region as the key input of the screen surface operation state reconstruction and subsequent automatic diagnosis.

[0100] Therefore, the embodiment can realize accurate positioning and risk level division of the abnormal region of the screen surface through steps S310-S320. By introducing the texture complexity weight, the model is not only sensitive to the deviation of particle distribution, but also can dynamically adjust the abnormal confidence according to the image details, greatly improving the anti-interference and accuracy of abnormal screening, and greatly enhancing the practical value of abnormal diagnosis.

[0101] In one embodiment, the screen surface defect judgment threshold includes a first abnormal saliency threshold, a second abnormal saliency threshold, and a morphological stability judgment threshold.

[0102] Step S400: generating a screen surface defect judgment result according to the saliency weight and the preset screen surface defect judgment threshold, including:

[0103] Step S410: constructing a classification function according to the saliency weight and the first abnormal saliency threshold, the second abnormal saliency threshold, and the morphological stability judgment threshold.

[0104] Step S420: generating a screen surface defect judgment result according to the classification function.

[0105] In the embodiment, in order to solve the problem that the existing screen surface defects are mainly judged by simple threshold or classified by experience rule, and the morphological characteristics of the abnormal region are not analyzed in depth, resulting in inaccurate and poor adaptability of the defect type, the embodiment constructs a classification function according to the saliency weight and the first abnormal saliency threshold, the second abnormal saliency threshold, and the morphological stability judgment threshold; and then generates a screen surface defect judgment result according to the classification function.

[0106] The classification function adopts a defect classification model based on block correlation clustering density, and the model is as follows:

[0107]

[0108] wherein,

[0109] is the abnormality weight of the region, reflecting the abnormality degree of the region. is the shape stability of the region. is the first abnormality weight threshold and the second abnormality weight threshold , is set as the 80th percentile or above in the abnormal region , and Generally, the 95th percentile of the normal region in all samples is taken. is the shape stability threshold, and the contour coefficient of variation of the particle contour of all image regions is calculated, such as the ratio of the standard deviation to the mean value of the area and the perimeter, and the value range in the normal screening state is counted, and the median or the 70th percentile is taken as the threshold. Below the threshold indicates that the particle shape is consistent and the shape stability is high. is the final defect classification label.

[0110] The classification function determines the defect type according to the abnormality weight of the region and the shape stability of the particle. In the defect classification model, the shape stability of the region is calculated based on the coefficient of variation of the particle contour, reflecting the regularity and consistency of the particle shape; specifically, it can be obtained by calculating the coefficient of variation of the area of all identified particle contours in the region, that is, the standard deviation of the particle area is divided by the average value, and the smaller the value, the more consistent the particle shape and the more regular the distribution.

[0111] abnormality weight threshold , is used to distinguish the severity of the abnormality; in the final defect classification label , 1 represents accumulation blockage, 2 represents partial load dispersion, and 3 represents a normal region.

[0112] The classification function determines the defect type according to the abnormality weight of the region and the shape stability of the particle. The region with high abnormality weight and low shape stability is determined as accumulation blockage (type 1), indicating that the material is accumulated and the shape is irregular; the region with high abnormality weight and high shape stability is determined as partial load dispersion (type 2), reflecting that the material is offset but the particle distribution is regular; the region with low abnormality weight is classified as normal (type 3). This dual-index discrimination effectively avoids misjudgment caused by a single index, and enhances the accuracy and stability of defect identification.

[0113] In this embodiment, the method is also provided with a continuous time sequence frame analysis mechanism, which takes the change process of the screening surface image in a time window into the analysis logic, is used to judge the dynamic stability and trend evolution characteristics of the screening state, so as to realize the early warning of "potential failure precursor".

[0114] The continuous time frame analysis mechanism is not to determine the abnormal degree of a single frame image, but to determine whether the evolution trend of abnormality exists a continuous amplification trend. To this end, in a continuous time period, the system records and analyzes the distribution form of the saliency region, the number of defect labels, and the spatial migration track of the center point of the abnormal region. The specific methods include:

[0115] First, trend variation judgment: by recording the change of abnormal labels of the same region in continuous images, if the abnormal labels continuously increase, it is determined as potential instability evolution.

[0116] Second, spatial drift feature extraction: track the moving track of the center point of each abnormal region over time, and judge whether there is a continuous trend of concentrating towards the edge or the middle of the screen surface.

[0117] Then, dynamic intervention mechanism suggestion output: based on the dynamic stability result, the system proposes adjustment suggestions, such as reducing the feeding rate, temporarily suspending the vibration module, checking the response of the unbalance sensor, etc.

[0118] Compared with the traditional image analysis system which mostly focuses on static judgment, the present application has time sequence trend recognition and response ability; it can realize early detection of slight but continuously intensified screen surface faults, such as small hole diffusion evolution into large area accumulation; it builds a closed loop logic path between image recognition results and control instructions, which is helpful for the subsequent formation of intelligent control system.

[0119] Therefore, the present embodiment realizes intelligent recognition and grade division of various screen defect types through inter-regional association clustering. This not only improves the accuracy of defect classification, but also effectively distinguishes different fault modes such as accumulation blockage and unbalance dispersion, provides more targeted basis for automatic adjustment and maintenance decision of the screening process, and significantly improves the intelligent level and practical value of the system.

[0120] In summary, the machine vision-based screen separation detection method and system provided in the present application can deploy an image acquisition component based on a screen light adaptation model, perform texture enhancement processing on the screen image collected by the image acquisition component, and obtain a texture enhancement factor, so as to adjust the image acquisition condition by constructing a light adaptation model, ensure the light balance and detail definition of the screen image, and significantly improve the image quality and the robustness of subsequent analysis. The particle distribution density is extracted according to the texture enhancement factor, and a global separation consistency index of the entire screen state is constructed according to the particle distribution density. The saliency weight is generated according to the global separation consistency index. The screen defect judgment result is generated according to the saliency weight and a preset screen defect judgment threshold. The multi-scale texture enhancement and dynamic block mechanism can strengthen the screen texture features in a complex particle motion background, effectively improve the particle boundary recognition capability, construct the screen separation consistency index based on the extracted particle distribution density, break the traditional isolated judgment mode of local anomalies, and realize the quantitative description of the global state of the screen. In addition, the present application also combines a local anomaly saliency evaluation model and a classification function to finely identify and intelligently classify the defect types of the screen separation abnormal area, which not only improves the accuracy and stability of the anomaly detection, but also effectively distinguishes different screen abnormal types such as blockage and unbalanced load, and solves the problems of lack of detailed classification of anomalies and high misjudgment rate in the prior art.

[0121] In one embodiment, as shown in Figure 2 a machine vision-based screen separation detection system is also provided, and the system comprises:

[0122] a texture enhancement factor generation module configured to deploy an image acquisition component based on a screen light adaptation model, and perform texture enhancement processing on the screen image collected by the image acquisition component to obtain a texture enhancement factor;

[0123] a consistency index generation module configured to extract a particle distribution density according to the texture enhancement factor, and construct a global separation consistency index of the entire screen state according to the particle distribution density;

[0124] a saliency weight generation module configured to generate a saliency weight according to the global separation consistency index;

[0125] a defect judgment result generation module configured to generate a screen defect judgment result according to the saliency weight and a preset screen defect judgment threshold.

[0126] In one embodiment, the texture enhancement factor generation module is further configured to:

[0127] The screen surface light adaptation model in the light reflection condition is constructed, the light source angle and intensity of the image acquisition assembly are adjusted according to the model, and the image acquisition assembly is deployed; the screen surface image collected by the image acquisition assembly is dynamically divided and subjected to multi-scale texture enhancement processing, and a texture enhancement factor is obtained.

[0128] In another embodiment, the consistency index generation module is further configured to: set a dynamic texture weight model according to the texture enhancement factor, construct a quantifiable image index for a separation state, and extract a particle distribution density; and construct a global separation consistency index of the entire screen surface state based on the extracted particle distribution density.

[0129] In another embodiment, the saliency weight generation module is further configured to: construct a local abnormal saliency evaluation model according to the global separation consistency index; and output a saliency weight based on the local abnormal saliency evaluation model.

[0130] In another embodiment, the saliency weight generation module is further configured to: identify a region with a serious separation abnormality in the screen surface image based on the saliency weight.

[0131] In another embodiment, the screen surface defect judgment threshold includes a first abnormal saliency threshold, a second abnormal saliency threshold, and a morphological stability judgment threshold; and the defect judgment result generation module is further configured to: construct a classification function according to the saliency weight and the first abnormal saliency threshold, the second abnormal saliency threshold, and the morphological stability judgment threshold; and generate a screen surface defect judgment result according to the classification function.

[0132] In one embodiment, as shown in Figure 3 The computer device further includes a system bus, an internal memory, a network structure, a display screen, an input device, and the like.

[0133] In one embodiment, a computer readable storage medium having a computer program stored thereon is provided, and the computer program is executed by a processor to implement the steps of the above-mentioned machine vision-based screen surface separation detection method.

[0134] It should be noted that the information interaction and execution process between the above-mentioned modules, since the same concept as the method embodiment of the present application, the specific functions and the technical effects brought about, specific can refer to the method embodiment part, here will not be repeated.

[0135] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit. In addition, the specific names of the functional units and modules are only for the convenience of mutual distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0136] It should be noted that the information interaction, execution process and the like between the above modules are based on the same concept as the method embodiments of the present application, and the specific functions and technical effects brought by them can be referred to the method embodiments part. Here, it will not be repeated.

[0137] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit. In addition, the specific names of the functional units and modules are only for the convenience of mutual distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0138] The embodiments of the present application also provide a network device, which comprises at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any of the above method embodiments when executing the computer program.

[0139] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program, wherein the computer program is executable by a processor to implement the steps in any of the above method embodiments.

[0140] The embodiments of the present application provide a computer program product, which, when running on a mobile terminal, causes the mobile terminal to perform the steps in the above-mentioned various method embodiments.

[0141] The integrated unit, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the present application can implement all or part of the processes in the above-mentioned embodiments by a computer program to instruct related hardware to complete, and the computer program can be stored in a computer readable storage medium. The computer program, when executed by a processor, can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium at least includes any entity or device capable of carrying the computer program code to the photographing device / terminal equipment, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium. For example, U disk, mobile hard disk, magnetic disk or optical disk, etc. In some jurisdictions, according to legislation and patent practice, the computer readable medium can not be an electrical carrier signal and a telecommunication signal.

[0142] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0143] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0144] In the embodiments of the present application, it should be understood that the disclosed apparatus / network device and method can be implemented in other manners. For example, the described apparatus / network device embodiments are merely illustrative. For example, the division of the modules or units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0145] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0146] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some 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 the embodiments of the present application, and should be included in the protection scope of the present application.

[0147] An embodiment of the present application further provides a computer device, which comprises at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the steps in any of the above-described embodiments.

[0148] The computer device can include, but is not limited to, a processor, a memory. Those skilled in the art can understand that the above description is an example of the computer device, and does not constitute a limitation on the computer device, and can include more or fewer components than the above description, or combine some components, or different components, for example, can also include input / output devices, network access devices, etc.

[0149] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0150] The memory can be an internal storage unit of the computer device in some embodiments, for example, a hard disk or a memory of the computer device. The memory can also be an external storage device of the computer device in other embodiments, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory can include both the internal storage unit and the external storage device of the computer device. The memory is used to store an operating system, an application program, a boot loader, data, and other programs, for example, program codes of the computer program, etc. The memory can also be used to temporarily store data that has been output or will be output.

[0151] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present disclosure.

[0152] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, however, it should not be understood as a limitation on the scope of the present application. It should be pointed out that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A machine vision based separation detection method for a screen deck of a separation machine, characterized by, The method comprises: deploying an image acquisition component based on a screen surface light illumination adaptability model, and performing texture enhancement processing on a screen surface image acquired by the image acquisition component to obtain a texture enhancement factor; extracting a particle distribution density according to the texture enhancement factor, and constructing a global separation consistency index of the entire screen surface state according to the particle distribution density; generating a saliency weight according to the global separation consistency index; generating a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold; deploying an image acquisition component based on a screen surface light illumination adaptability model, and performing texture enhancement processing on a screen surface image acquired by the image acquisition component to obtain a texture enhancement factor, comprising: constructing a screen surface light illumination adaptability model under the condition of light reflection, and adjusting the light source angle and intensity of the image acquisition component according to the model to deploy the image acquisition component; performing dynamic block and multi-scale texture enhancement processing on the screen surface image acquired by the image acquisition component to obtain a texture enhancement factor; extracting a particle distribution density according to the texture enhancement factor, and constructing a global separation consistency index of the entire screen surface state according to the particle distribution density, comprising: setting a dynamic texture weight model according to the texture enhancement factor, constructing a quantifiable image index for the separation state, and extracting a particle distribution density; constructing a global separation consistency index of the entire screen surface state based on the extracted particle distribution density; generating a saliency weight according to the global separation consistency index, comprising: constructing a local anomaly saliency evaluation model according to the global separation consistency index; outputting a saliency weight based on the local anomaly saliency evaluation model; The screen surface light illumination adaptability model is as follows: ； wherein, is the screen illumination adaptability, is the average gray value of the unit area screen image, which is calculated by statistically averaging the pixel gray value in the unit area, and then using the gray scale or converting the color image to a gray scale image, for reflecting the overall brightness level of the region; is the image edge definition coefficient, which is calculated by statistically averaging the average intensity of the edge response value in the unit area through edge detection of the image by Sobel or Canny operator, is the background noise ratio, which is calculated by comparing the standard deviation of the background area with the gray value of the main area of the image, is a constant set to prevent the denominator from being zero; The global separation consistency index is as follows: ； wherein, is a variance-weighted consistency evaluation index of particle distribution density, i.e., a global separation consistency index, is a particle distribution density, is an average value of all , represents the number of regions into which the image is divided in the vertical direction, N represents the number of regions into which the image is divided in the horizontal direction, and M*N is the total number of regions into which the image is divided. The dynamic texture weight model is as follows: ； wherein, is a texture enhancement factor, is the number of particle identifications within the region, is the area of the region.

2. The machine vision-based separator screen deck separation detection method of claim 1, wherein, The screen surface defect judgment threshold comprises a first anomaly saliency threshold, a second anomaly saliency threshold, and a morphological stability judgment threshold; generating a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold, comprising: constructing a classification function according to the saliency weight and the first anomaly saliency threshold, the second anomaly saliency threshold, and the morphological stability judgment threshold; generating a screen surface defect judgment result according to the classification function.

3. A machine vision based separation machine screen separation detection system, characterized by, The system comprises: a texture enhancement factor generation module for deploying an image acquisition component based on a screen surface light illumination adaptability model, and performing texture enhancement processing on a screen surface image acquired by the image acquisition component to obtain a texture enhancement factor; a consistency index generation module for extracting a particle distribution density according to the texture enhancement factor, and constructing a global separation consistency index of the entire screen surface state according to the particle distribution density; a saliency weight generation module for generating a saliency weight according to the global separation consistency index; a defect judgment result generation module for generating a screen surface defect judgment result according to the saliency weight and a preset screen surface defect judgment threshold.

4. The machine vision-based separator screen deck separation detection system of claim 3, wherein, The texture enhancement factor generation module is further configured to: The screen surface light adaptation model in the construction of light reflection condition is used to adjust the light source angle and intensity of the image acquisition component to deploy the image acquisition component; the screen surface image collected by the image acquisition component is subjected to dynamic block and multi-scale texture enhancement processing to obtain a texture enhancement factor. 5.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-4 when the computer program is executed by the processor. The processor implements the steps of the method of claim 1 or 2 when executing the computer program.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of claim 1 or 2.

Images