Plastic particle detection method and system based on image recognition

Abstract

The invention discloses a plastic particle detection method and system based on image recognition, and relates to the technical field of plastic particle detection, and the method comprises the steps: obtaining plastic particle image data, and obtaining particle weight sensor actual measurement data corresponding to the image data; constructing an image recognition model based on the image data to obtain an image estimation total weight; based on the image estimation total weight and actual measurement data of the weight sensor, constructing a feedback verification model and performing calculation to obtain an artifact suspected degree value; and judging artifact interference in the image recognition model according to the image estimation total weight and the artifact suspected degree value, thereby realizing plastic particle detection based on image recognition. The problems that in the prior art, image recognition errors are difficult to correct, and artifact interference cannot be quantized are effectively solved, the detection accuracy of the detection method in a complex environment is improved, and cooperative control of an image detection system and a feedback device is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of plastic particle detection, and particularly to a plastic particle detection method and system based on image recognition. Background Art

[0002] In recent years, during the production, sorting, and packaging of plastic particles, parameters such as the quantity, shape, color difference, and quality consistency of the particles have an important impact on product performance and subsequent processing. With the improvement of industrial automation levels, more and more enterprises use image recognition technology to conduct on-line detection of plastic particles to replace the traditional manual visual inspection method and achieve efficient and non-contact quality monitoring. Existing systems mostly collect images through industrial cameras and combine image recognition algorithms to judge the appearance characteristics of the particles.

[0003] Currently, the detection methods used in the field of plastic particle detection are difficult to achieve dynamic determination and closed-loop correction of artifact interference during actual use. Especially when image misjudgment cannot be immediately feedback-calibrated, it is easy to cause inaccurate particle recognition quantity and deviation in quality assessment, thereby affecting overall process control and data credibility. Summary of the Invention

[0004] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title. However, such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the problems existing in the above-mentioned existing plastic particle detection method and system based on image recognition, the present invention is proposed.

[0006] Therefore, the purpose of the present invention is to provide a plastic particle detection method and system based on image recognition, which is suitable for solving the problems that are prone to cause inaccurate particle recognition quantity and deviation in quality assessment, thereby affecting overall process control and data credibility.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, an embodiment of the present invention provides a plastic particle detection method based on image recognition, including: Obtain plastic particle image data and obtain the actual measurement data of the particle weight sensor corresponding to the image data; Build an image recognition model based on the image data to obtain an estimated total weight of the image; Build a feedback verification model based on the estimated total weight of the image and the actual measurement data of the weight sensor and perform calculations to obtain an artifact suspicion degree value; Estimate the total weight and the degree of artifact suspicion value according to the image to determine the artifact interference in the image recognition model, and realize the plastic particle detection based on image recognition.

[0008] As a preferred embodiment of the plastic particle detection method based on image recognition according to the present invention, wherein: the image data refers to the plastic particle image data collected by an image acquisition device, and the image data includes the distribution, color, shape, edge contour, projected area and image gray information of the particles in the image; The actual measured data of the weight sensor refers to the total weight data of the current batch of plastic particles collected in real time by a quality detection device arranged on the particle conveying path or the weighing platform.

[0009] As a preferred embodiment of the plastic particle detection method based on image recognition according to the present invention, wherein: constructing an image recognition model based on the image data to obtain an estimated total weight of the image, including the following steps: By introducing color saturation, edge gradient, image gray level and shape parameters, perform non-linear combination processing on the original pixel features of each plastic particle in the image, construct an image feature enhancement function for a single particle, and obtain an image feature response value, so as to comprehensively describe the key recognition features of each plastic particle in the image. The specific formula is as follows: ; Wherein, is the image feature response value of the i-th particle, represents the color saturation of the i-th particle, represents the edge gradient intensity, represents the gray mean value of the particle area, represents the particle roundness ratio; by introducing a power function and a proportional mapping factor, perform non-linear transformation on the image feature response value to obtain the estimated weight of a single particle image, so as to enhance the accuracy and stability of the image recognition model in estimating the particle mass under the conditions of drastic changes in particle size or inconsistent image scales. The specific formula is as follows: ; Wherein, is the estimated weight of the i-th particle image, is the standard density of the plastic particles, is the proportional adjustment factor; Construct an image recognition model according to the optimized estimated weight of a single particle image to obtain the result of the estimated total weight of the image.

[0010] As a preferred embodiment of the plastic particle detection method based on image recognition according to the present invention, wherein: the formula of the image recognition model is as follows: ; Among them, is the estimated total weight of the image, where n is the number of particles identified in the image.

[0011] As a preferred solution of the plastic particle detection method based on image recognition according to the present invention, wherein: based on the estimated total weight of the image and the measured data of the weight sensor, a feedback verification model is constructed and calculated to obtain an artifact suspicion degree value, including the following steps: By introducing the hyperbolic tangent function and the logarithmic function to process the composite effect of image estimation error and particle distribution fluctuation, a normalized response score is obtained, thereby enhancing the response ability of the feedback verification model to scenarios of uneven particle distribution or sudden changes in image recognition deviation. The specific formula is as follows: ; Among them, is the normalized response score, is the measured total weight of plastic particles by the weight sensor, is the normalized adjustment constant, is the variance of the image particle distribution; by introducing an exponential decay function to process the relative change rate of the measured weight within the current time period, a feedback verification model is constructed, thereby suppressing the misjudgment effect caused by short-term fluctuations during the particle transportation process, and obtaining an artifact suspicion degree value. The specific formula of the feedback verification model is as follows: ; Among them, is the artifact suspicion degree value, is the suppression adjustment factor, which is used to control the sensitivity of the model to the real weight fluctuation, is the measured weight at the current time point, is the measured weight at the previous time point.

[0012] As a preferred solution of the plastic particle detection method based on image recognition according to the present invention, wherein: the determination states of the artifact interference include no interference, acceptable interference, moderate artifact interference, and severe artifact interference.

[0013] As a preferred solution of the plastic particle detection method based on image recognition according to the present invention, wherein: judging the artifact interference in the image recognition model according to the estimated total weight of the image and the artifact suspicion degree value includes the following steps: Based on the absolute error between the estimated total weight of the image and the measured data of the weight sensor, a basic interference judgment is made. If the absolute error is less than or equal to the first error threshold W0 and the artifact suspicion degree value is less than or equal to the first interference threshold M0, it is determined that the current state is no interference, and the plastic particle detection based on image recognition is continued; If the absolute error is less than or equal to the first error threshold W0 and the artifact suspicion degree value is greater than the first interference threshold M0, preliminarily determine that the current state is acceptable interference and execute the next step; When the preliminary determination result is acceptable interference, perform interference escalation judgment based on the distribution uniformity coefficient of the image feature response value and the dispersion of the estimated weight of single-particle images. If the distribution uniformity coefficient is greater than or equal to the uniformity threshold A0 and the dispersion is less than or equal to the dispersion threshold B0, determine that the current state is acceptable interference, send a first-level prompt message to the operator, and suggest performing an image sensor cleaning inspection; If the distribution uniformity coefficient is less than the uniformity threshold AO or the dispersion is greater than the dispersion threshold B0, determine that the current state is moderate artifact interference, send a second-level warning message to the operator, and immediately trigger the parameter calibration process of the image recognition model; If the absolute error is greater than the first error threshold W0 and less than or equal to the second error threshold W0', and the artifact suspicion degree value is less than or equal to the second interference threshold M0', preliminarily determine that the current state is moderate artifact interference and execute the next step; When the preliminary determination result is moderate artifact interference, perform interference escalation judgment based on the proportion of the abnormal gray area in the image and the number of abnormal particles in the edge contour. If the proportion of the abnormal gray area is less than or equal to the area threshold C0 and the number of abnormal particles in the edge contour is less than or equal to the number threshold D0, determine that the current state is moderate artifact interference, send a second-level warning message to the operator, and immediately trigger the parameter calibration process of the image recognition model; If the proportion of the abnormal gray area is greater than the area threshold C0 or the number of abnormal particles in the edge contour is greater than the number threshold D0, determine that the current state is severe artifact interference, send a third-level alarm message, stop the current detection process and start the standby detection system; If the absolute error is greater than the second error threshold W0' or the artifact suspicion degree value is greater than the second interference threshold M0', determine that the current state is severe artifact interference, send a third-level alarm message, stop the current detection process and start the standby detection system.

[0014] In a second aspect, in order to further solve the problem that it is easy to cause inaccurate particle recognition quantity and quality evaluation deviation, which in turn affects the overall process control and data credibility, the present embodiment provides a plastic particle detection system based on image recognition, including: A data acquisition module for acquiring plastic particle image data and acquiring the measured data of the particle weight sensor corresponding to the image data; An identification model construction module: for constructing an image recognition model based on the image data to obtain the estimated total weight of the image; Calibration model construction module: used to construct a feedback calibration model and perform calculations based on the estimated total weight from the image and the measured data of the weight sensor, and obtain an artifact suspicion degree value; Intelligent detection module: used to determine the artifact interference in the image recognition model according to the estimated total weight from the image and the artifact suspicion degree value, and realize the plastic particle detection based on image recognition.

[0015] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, and the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the plastic particle detection method based on image recognition as described in the first aspect of the present invention is implemented.

[0016] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by the processor, any step of the plastic particle detection method based on image recognition as described in the first aspect of the present invention is implemented.

[0017] Advantages of the present invention: By combining the image recognition model with the feedback data of the weight sensor, the present invention establishes an identification algorithm for extracting multi-parameter image features including color saturation, edge gradient, image gray scale, etc., realizes the high-precision mapping and error evaluation between the estimated weight and the actual weight of the particle image, and can quickly perform abnormal level judgment and system response when an identification error occurs by constructing an artifact suspicion degree value and a grading interference determination mechanism, effectively solving the problems in the prior art that it is difficult to correct image recognition errors and quantify artifact interference, improving the detection accuracy of the detection method in complex environments. At the same time, the present invention can realize the coordinated control of the image detection system and the feedback device, significantly improving the overall detection ability without increasing the hardware cost, and having good engineering promotion value; The present invention is applicable to the on-line identification, quality determination and abnormal interference detection of plastic particles, especially applicable to image detection in scenarios such as particle transportation, sorting or packaging, and can be used for real-time analysis of the number, morphological integrity and estimated weight of particles in an automated production line. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings according to these drawings without creative efforts. Among them: Figure 1 It is a schematic diagram of the overall process of the plastic particle detection method based on image recognition proposed by the present invention; Figure 2Schematic diagram of the artifact interference determination logic for the plastic particle detection method based on image recognition proposed by the present invention. Detailed implementation manners

[0019] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the detailed implementation manners of the present invention with reference to the accompanying drawings of the specification.

[0020] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0021] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other from other embodiments.

[0022] Thirdly, the present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention in detail, for the sake of clarity, the cross-sectional views showing the device structure will be enlarged locally out of the general scale, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0023] Embodiment 1 Refer to Figure 1 - Figure 2 , which is an embodiment of the present invention, and provides a plastic particle detection method based on image recognition.

[0024] The existing plastic particle detection methods based on image recognition mainly have the following problems: it is difficult to achieve dynamic determination and closed-loop correction of artifact interference. Especially when image misjudgment cannot be immediately feedback-calibrated, it is easy to cause inaccurate particle recognition quantity and deviation in quality assessment, thereby affecting the overall process control and data credibility.

[0025] This application provides a solution that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to elaborate in detail how to implement the plastic particle detection method based on image recognition.

[0026] Figure 1 shows the overall flowchart of the plastic particle detection method based on image recognition, including: S1: Obtain plastic particle image data and obtain the measured data of the particle weight sensor corresponding to the image data.

[0027] The image data refers to the plastic particle image data collected by an image acquisition device. Here, the image acquisition device can be a vision sensing device with imaging capabilities, such as an industrial camera, a line scan camera, or a multispectral imaging device, which is used to obtain information including the distribution, color, shape, edge contour, projected area, and image grayscale of the particles in the image. The actual measured data of the weight sensor refers to the total weight data of the current batch of plastic particles collected in real time by a mass detection device set on the particle conveying path or the weighing platform. Here, the mass detection device can be a physical quantity acquisition device with real-time weight detection capabilities, such as a static or dynamic weighing sensor, a piezoelectric load cell, and an electromagnetic force balance type mass determination module, etc.

[0028] This step can lay a high-precision basic data support for the construction of the subsequent image recognition model and the setting of the feedback verification model by fusing the image data collected by the vision sensing device and the actual measured weight data collected by the mass detection device, ensuring the mapping consistency in the spatial dimension and the complementary nature in the feature dimension between the particle image features and the physical quality information.

[0029] S2: Construct an image recognition model based on the image data to obtain the estimated total weight of the image.

[0030] Furthermore, constructing an image recognition model based on the image data to obtain the estimated total weight of the image includes the following steps: By introducing color saturation, edge gradient, image grayscale, and shape parameters, perform non-linear combination processing on the original pixel features of each plastic particle in the image, construct an image feature enhancement function for a single particle, and obtain an image feature response value, thereby comprehensively depicting the key recognition features of each plastic particle in the image. The specific formula is as follows: ; Where, is the image feature response value of the i-th particle, represents the color saturation of the i-th particle, represents the edge gradient intensity, represents the average grayscale of the particle area, represents the roundness ratio of the particle; Furthermore, according to the obtained image feature response values, calculate the distribution uniformity coefficient of the image feature response values; First, count the image feature response values of all particles in the image ; Calculate the mean value and the standard deviation ; The specific formula is as follows: ; Among them, A is the distribution uniformity coefficient, and the smaller it is, the more uniform the distribution is.

[0031] By introducing a power function and a proportional mapping factor to perform a non-linear transformation on the image feature response value, the estimated weight of a single-particle image is obtained, thereby enhancing the accuracy and stability of the image recognition model in estimating the particle mass under the conditions of drastic changes in particle size or inconsistent image scales. The specific formula is as follows: ; Among them, is the estimated weight of the image of the i-th particle, is the standard density of the plastic particles, is the proportional adjustment factor; Further, the dispersion of the estimated weight of a single-particle image is calculated according to the estimated weight of the image. First, the estimated weights of the images of all particles are counted ; The specific formula is: ; Among them, IQR is the interquartile range of the weight obtained by the calculation method using the interquartile range, is the median of the weight, and B is the dispersion. The larger it is, the higher the dispersion degree of the weight estimation is.

[0032] An image recognition model is constructed based on the optimized estimated weight of a single-particle image to obtain the result of the estimated total weight of the image.

[0033] Further, the formula of the image recognition model is as follows: ; Among them, is the estimated total weight of the image, where n is the number of particles recognized in the image.

[0034] In the embodiment of the present application, in the construction process of the image recognition model, by introducing color saturation, edge gradient intensity, image grayscale, and shape parameters to describe the image response characteristics of plastic particles under different lighting and background conditions, the ability of the image recognition model to distinguish particles with blurred contours and similar colors can be effectively enhanced. The proportional adjustment factor can be obtained by comparing the two-dimensional projected area of the particles in the image with the known particle volume data for training, or can be calibrated by setting a reference scale or known-size standard particles, and is used to realize the mapping relationship between the image scale and the actual physical size, ensuring the consistency of the model output under different image resolutions or shooting angles.

[0035] It should be noted that traditional image recognition methods mostly adopt linear regression or average pixel gray-scale estimation methods, which are difficult to fully extract the fine-grained image differences of particles, resulting in a high misrecognition rate and large estimation deviation in complex backgrounds; the embodiments of the present application break through the limitations of the insufficient image feature extraction ability of traditional methods by constructing an image recognition model that integrates an image enhancement function and a non-linear weighted mapping, significantly improving the correlation between the detection accuracy of plastic particles in the image and the estimated weight, and providing reliable technical support for realizing real-time and high-precision online particle detection.

[0036] S3: Based on the estimated total weight of the obtained image and the measured data of the weight sensor, construct a feedback verification model and perform calculations to obtain the degree of suspicion of artifacts.

[0037] Based on the estimated total weight of the obtained image and the measured data of the weight sensor, constructing a feedback verification model and performing calculations to obtain the degree of suspicion of artifacts includes the following steps: By introducing the hyperbolic tangent function and the logarithmic function to process the composite effect of the image estimation error and the particle distribution fluctuation, a normalized response score is obtained, thereby enhancing the response ability of the feedback verification model to scenarios of uneven particle distribution or sudden changes in image recognition deviation. The specific formula is as follows: ; Where, is the normalized response score, is the total weight of the plastic particles actually measured by the weight sensor, is the normalized adjustment constant, is the variance of the image particle distribution; By introducing the exponential decay function to process the relative change rate of the actually measured weight within the current time period, a feedback verification model is constructed to suppress the misjudgment effect caused by short-term fluctuations during the particle transportation process, and the degree of suspicion of artifacts is obtained. The specific formula of the feedback verification model is as follows: ; Where, is the degree of suspicion of artifacts, is the suppression adjustment factor used to control the sensitivity of the model to the real weight fluctuation, is the actually measured weight at the current time point, is the actually measured weight at the previous time point.

[0038] Furthermore, calculate the proportion of the abnormal gray-scale area of the image according to the obtained degree of suspicion of artifacts; First, perform a gray-scale histogram analysis on the image and set the normal gray-scale range ; Statistical area of the pixel region where the gray-scale value exceeds this range and the total particle projection area The ratio is as follows: ; Among them, C is the proportion of the abnormal gray area of the image.

[0039] It should be noted that the number of abnormal particles with edge contours is determined as follows: The image recognition module is used to segment each plastic particle in the image and extract its edge contour pixel area. Specifically, it is preferably to use the Canny edge detection algorithm or Sobel operator to extract the edge gradient to obtain the binary boundary image of the particle contour; fit and analyze each extracted particle edge contour, and calculate the feature indexes such as the closure degree, edge continuity, number of edge corner points, and edge gradient change rate of the particle contour. Preferably, if any one of the following three judgment conditions is met, the particle is considered an abnormal particle with an edge contour: The contour closure degree is lower than the set threshold (for example, the closure degree < 85%), indicating that the edge is incomplete or there is a break; The number of edge corner points exceeds the set upper limit (for example, > 10 acute angle points), indicating that there are obvious sawteeth or damages on the particle surface; The edge gradient change rate fluctuates abnormally violently in the boundary area. For example, its standard deviation exceeds twice the average value of normal particles, indicating that there may be chipping, fragmentation or artifact interference on the edge.

[0040] The system counts the number of particles that meet any of the above conditions among all the detected particles as the number of abnormal particles with edge contours, and uses this value as one of the important indexes for subsequent evaluation of the overall quality of the particles or judgment of the degree of artifact interference.

[0041] Exemplarily, in a certain plastic particle detection scenario, the system collected and calculated a frame of image as follows. The image recognition module recognized a total of 50 particles, and estimated the total weight to be: , and the weighing sensor actually measured the weight of the current batch of plastic particles as , the measured weight at the previous time point (for example, the previous 0.5s) was 236.7g, and the variance of the number of particles in this image frame counted by the image recognition model was: , set the normalization adjustment constant to , set the fluctuation suppression adjustment coefficient ; Then the image error normalization function value is ; The suspected degree value of artifacts . Assume that in this detection process, the first interference threshold M0 is set to 0.3, then it can be determined that the image recognition result is not infected and the recognition result is credible.

[0042] In an embodiment of the present application, the normalized adjustment constant can be determined by constructing an error function model between the image estimated weight and the measured weight, and performing parameter fitting with the goal of minimizing the normalized error deviation, thereby determining the normalized adjustment constant that best suits the current particle image feature distribution; this parameter can also be set as a fixed constant in combination with the density fluctuation range of different types of plastic particles in actual applications, to avoid the problem of the error denominator approaching zero under extreme working conditions; the inhibition adjustment factor can be obtained by fitting and analyzing the measured weight change rate per unit time with known stable state data, constructing an error attenuation function, and solving it by minimizing the misjudgment risk function; it can also be set as an empirical parameter in the system design stage in combination with the vibration characteristics of the particle conveyor line and the stability index of the weighing platform, to adjust the sensitive response of the exponential attenuation function to weight fluctuations.

[0043] It should be noted that traditional image detection methods usually separate the particle recognition process from the image quality assessment process, lacking a deep tracing mechanism for the recognition distortion caused by image interference factors (such as edge blur, ghosting, water stain reflection, etc.), resulting in the system being unable to effectively identify potential misjudgment trends at the early stage of the abnormality. This embodiment introduces an image edge contour abnormality recognition mechanism, constructs an abnormal particle quantity statistical function based on local edge quality indicators, and incorporates the statistical results into the comprehensive judgment process of the artifact interference state, which can accurately capture the recognition structure abnormality caused by local interference in the image; together with the proportion of image grayscale abnormal areas, it constitutes an important reference dimension for image stability, enabling the system to achieve quantitative evaluation of image detail level disturbances, thereby sending intervention signals before the artifacts spread over a large area. This method breaks through the limitation of traditional particle recognition algorithms on global contour dependence, improves the system's response sensitivity to tiny boundary disturbances, provides basic support for the dynamic self-calibration of online image recognition models, and enhances the adaptability of the detection method in complex background environments.

[0044] S4: Determine the artifact interference in the image recognition model based on the estimated total weight of the image and the artifact suspicion level value, and realize plastic particle detection based on image recognition.

[0045] Preferably, the determination status of artifact interference includes no interference, acceptable interference, moderate artifact interference and severe artifact interference.

[0046] Specifically, Figure 2 As shown, determining the artifact interference in the image recognition model according to the image estimated total weight and the artifact suspicion degree value includes the following steps: A basic interference judgment is performed based on the absolute error between the total weight estimated by the image and the actual measured data of the weight sensor. If the absolute error is less than or equal to the first error threshold W0 and the artifact suspicion level value is less than or equal to the first interference threshold M0, the current state is determined to be free of interference, and the plastic particle detection based on the image recognition model continues; If the absolute error is less than or equal to the first error threshold W0 and the artifact suspicion degree value is greater than the first interference threshold M0, then preliminarily determine that the current state is acceptable interference and execute the next step; When the preliminary determination result is acceptable interference, perform interference escalation judgment based on the distribution uniformity coefficient of the image feature response value and the dispersion of the estimated weight of single-particle images. If the distribution uniformity coefficient is greater than or equal to the uniformity threshold A0 and the dispersion is less than or equal to the dispersion threshold B0, then determine that the current state is acceptable interference, and send a first-level prompt message to the operator, suggesting to conduct an image sensor cleaning inspection; If the distribution uniformity coefficient is less than the uniformity threshold AO or the dispersion is greater than the dispersion threshold B0, then determine that the current state is moderate artifact interference, send a second-level warning message to the operator, and immediately trigger the parameter calibration process of the image recognition model; The parameter calibration process is as follows: The system will trigger the image data cache of the current image frame and several adjacent frames before and after (such as ±2 frames), and extract all the basic image features of the plastic particles therein, including but not limited to: the distribution, color, shape, edge contour, projected area, and image gray information of the particles in the image; Compare the above image features with the normal recognition parameter range initially set by the system to identify the feature items with a deviation trend; Based on the above deviation features, automatically adjust the recognition parameters or weight factors corresponding to the feature in the image recognition model, for example: enhancing the edge response of low-contrast images, correcting the particle segmentation threshold, etc.; The adjusted image recognition model re-recognizes the current image and evaluates it according to the artifact suspicion degree value; If the artifact suspicion degree value of the recognition result after calibration is significantly reduced, then save the current parameter settings as the model's temporary running configuration, otherwise return to the state before calibration and prompt for manual re-inspection.

[0047] If the absolute error is greater than the first error threshold W0 and less than or equal to the second error threshold W0', and the artifact suspicion degree value is less than or equal to the second interference threshold M0', then preliminarily determine that the current state is moderate artifact interference and execute the next step; When the preliminary determination result is moderate artifact interference, perform interference escalation judgment based on the proportion of the abnormal gray area in the image and the number of particles with abnormal edge contours. If the proportion of the abnormal gray area is less than or equal to the area threshold C0 and the number of particles with abnormal edge contours is less than or equal to the number threshold D0, then determine that the current state is moderate artifact interference, send a second-level warning message to the operator, and immediately trigger the parameter calibration process of the image recognition model; If the proportion of the abnormal grayscale area is greater than the area threshold C0 or the number of abnormal particles in the edge contour is greater than the quantity threshold D0, it is determined that the current state is severe artifact interference, a level-three alarm message is sent, the current detection process is stopped, and the standby detection system is started. If the absolute error is greater than the second error threshold W0' or the artifact suspicion degree value is greater than the second interference threshold M0', it is determined that the current state is severe artifact interference, a level-three alarm message is sent, the current detection process is stopped, and the standby detection system is started.

[0048] Among them, starting the standby detection system includes the following steps: The main detection system writes the current image frame number, timestamp, abnormal indicators (abnormal grayscale ratio, number of abnormal edge particles), and model operation parameters into the abnormal record cache, suspends the current image recognition and result upload process, prevents incorrect results from entering the quality control chain, sends a level-three alarm message to the system management layer, and at the same time triggers the standby system initialization signal.

[0049] The standby detection system refers to a redundant image recognition path built into this system that is independent of the main detection process. It can include any one or more of the following implementation methods: switching to a standby industrial camera and auxiliary lighting components to collect images of the same detection area, or calling another image recognition model with a different main model structure (for example, the main model is YOLO, and the standby model is a combination of ResNet + FCN); by switching the combination of the "image input channel" and the "recognition algorithm channel", a rapid fault bypass reconstruction of the image recognition link is achieved.

[0050] After the standby detection system is started, the current batch of plastic particles is re-detected according to the same image acquisition and recognition logic as the main detection system. Specifically, it includes the following operations: starting the standby image acquisition channel (such as a standby industrial camera) to re-acquire images of the current area to be detected; inputting the newly acquired image data into the original image recognition model for processing; combining the image recognition results with the measured data of the real-time weight sensor, and performing interference judgment through the artifact suspicion degree value calculation module; if the artifact suspicion degree value is less than the set threshold, it indicates that the detection result is credible, and the system outputs the current standby detection result as valid recognition data; if the artifact suspicion degree value is still greater than the threshold, the current image is marked as a "high-risk interference image", and it enters the manual re-inspection process or the system standby mode.

[0051] Exemplarily, assume that on a plastic particle production line, an image recognition-based plastic particle detection system is running to monitor the quality of plastic particles on the conveyor belt. After running continuously for several hours, the system discovers through the image data collected by the industrial camera that the particle distribution shows a slight unevenness, and the absolute error between the estimated total weight of the image and the actual measured data of the weight sensor slightly increases, but still does not exceed the first error threshold. However, the suspected degree value of artifacts gradually increases and exceeds the first interference threshold. The system calculates according to the feedback verification model and detects that the uniformity coefficient of the distribution of the image feature response values is slightly lower than the uniformity threshold, triggering a preliminary determination of acceptable interference, and sending a first-level prompt message to the operator, suggesting to check whether there is dust or light interference in the image sensor. The operator then cleans the industrial camera lens and finds that there is slight contamination on the lens surface. After cleaning, the system returns to normal, the suspected degree value of artifacts decreases, and the error between the estimated total weight of the image and the measured data further shrinks, ensuring the accuracy of particle quality detection and avoiding misjudgment of particle quantity or quality assessment deviation caused by artifact interference.

[0052] In the embodiment of the present application, for each threshold in the artifact interference determination, the first error threshold and the second error threshold are determined by statistically analyzing the differences between the estimated total weights of a large number of images collected by the plastic particle detection system under normal operating conditions and the actual measured data of the weight sensor, calculating the standard deviation and confidence interval, and comprehensively evaluating in combination with the technical specifications and error tolerance ranges of the industrial camera and the weight sensor; the first interference threshold and the second interference threshold are initially set by the quantile method based on the statistical distribution characteristics of the suspected degree values of artifacts by the feedback verification model, by analyzing the artifact interference data in different scenarios (such as light changes, lens contamination, particle stacking, etc.), and are verified and adjusted through simulation tests; the uniformity threshold and the dispersion threshold are determined by long-term observation of the distribution uniformity coefficient of the image feature response values and the dispersion of the estimated weights of particle images, in combination with the distribution characteristics of particles on the conveyor belt and the stability analysis of the image recognition algorithm, aiming to effectively distinguish normal particle distribution fluctuations from abnormalities caused by artifact interference; the area threshold and the quantity threshold are set by comprehensively evaluating the distribution characteristics of the proportion of abnormally gray areas in the image and the number of abnormally contoured particles, in combination with common interference factors (such as particle overlap, background noise, etc.) in the actual production environment. In addition, the above thresholds can determine the initial values through small-scale test data in actual applications, and be optimized through continuous monitoring and feedback, or obtain a better threshold combination through training with historical data in combination with machine learning methods. This embodiment does not make specific limitations on this.

[0053] It should be noted that traditional plastic particle detection methods mostly rely on a single image recognition algorithm, lacking a dynamic determination and closed-loop correction mechanism for artifact interference. It is easy to cause recognition errors due to factors such as light changes, lens contamination, or particle stacking, affecting the accuracy of particle quantity and quality assessment. In this embodiment, a multi-level interference determination mechanism combining an image recognition model and a feedback verification model is constructed, establishing a refined determination system for no interference, acceptable interference, moderate artifact interference, and severe artifact interference. By comprehensively considering the estimated total weight of the image, the degree of artifact suspicion value, and the particle distribution characteristics, dynamic recognition and hierarchical response to artifact interference are achieved. This mechanism overcomes the limitation of single-threshold judgment in traditional detection methods and can take corresponding warning and calibration measures according to the nature, degree, and development trend of interference, significantly improving the robustness and accuracy of the detection system in complex production environments, reducing misjudgments caused by artifact interference, avoiding unnecessary downtime and maintenance costs, and optimizing the process control and data credibility of plastic particle detection. In summary, the present invention combines an image recognition model with the feedback data of a weight sensor to establish an identification algorithm for extracting multi-parameter image features including color saturation, edge gradient, image grayscale, etc., realizing a high-precision mapping and error evaluation between the estimated weight and the actual weight of the particle image. By constructing a degree of artifact suspicion value and a hierarchical interference determination mechanism, it can quickly perform abnormal level judgment and system response when recognition errors occur, effectively solving the problems in the prior art that image recognition errors are difficult to correct and artifact interference cannot be quantified, improving the detection accuracy of this detection method in complex environments. At the same time, the present invention can realize the coordinated control of the image detection system and the feedback device, significantly improving the overall detection ability without increasing hardware costs, and having good engineering promotion value.

[0054] Embodiment 2 This is an embodiment of the present invention, providing a plastic particle detection system based on image recognition, including: A data acquisition module for acquiring plastic particle image data and the measured data of the particle weight sensor corresponding to the image data; An identification model construction module for constructing an image recognition model based on the image data to obtain the estimated total weight of the image; A verification model construction module for constructing a feedback verification model and performing calculations based on the obtained estimated total weight of the image and the measured data of the weight sensor to obtain the degree of artifact suspicion value; An intelligent detection module for determining the artifact interference in the image recognition model according to the estimated total weight of the image and the degree of artifact suspicion value, realizing plastic particle detection based on image recognition.

[0055] Embodiment 3, which is an embodiment of the present invention and is different from the previous embodiment in that: If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0056] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or used in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0057] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), optical fiber devices, and portable compact disc read-only memories (CDROMs). Additionally, a computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or processing it in other suitable ways as necessary, and then storing it in a computer memory.

[0058] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following technologies well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application specific integrated circuits with suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0059] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting plastic particles based on image recognition, characterized in that, Including: Obtain plastic particle image data and obtain the actual measured data of the particle weight sensor corresponding to the image data; Construct an image recognition model based on the image data to obtain an estimated total weight of the image; Construct a feedback verification model based on the estimated total weight of the image and the actual measured data of the weight sensor and perform calculations to obtain an artifact suspicion degree value; Determine the artifact interference in the image recognition model according to the estimated total weight of the image and the artifact suspicion degree value, and realize the detection of plastic particles based on image recognition.

2. The plastic particle detection method based on image recognition according to claim 1, characterized in that: The image data refers to the plastic particle image data collected by an image acquisition device, and the image data includes the distribution, color, shape, edge contour, projection area and image gray information of the particles in the image; The actual measured data of the weight sensor refers to the total weight data of the current batch of plastic particles collected in real time by a quality detection device arranged on the particle conveying path or the weighing platform.

3. The method for detecting plastic particles based on image recognition according to claim 2, wherein: Construct an image recognition model based on the image data to obtain an estimated total weight of the image, including the following steps: By introducing color saturation, edge gradient, image gray level and shape parameters, perform non-linear combination processing on the original pixel features of each plastic particle in the image, construct an image feature enhancement function for a single particle, obtain an image feature response value, so as to comprehensively describe the key recognition features of each plastic particle in the image, and calculate the distribution uniformity coefficient of the image feature response value; Perform non-linear transformation on the image feature response value by introducing a power function and a proportional mapping factor to obtain the estimated weight of a single particle image, and calculate the dispersion of the estimated weight of a single particle image; Construct an image recognition model according to the optimized estimated weight of a single particle image to obtain the result of the estimated total weight of the image.

4. The method for detecting plastic particles based on image recognition according to claim 3, wherein: The image recognition model, the specific formula is as follows: ； wherein, is the estimated total weight of the image, where n is the number of particles identified in the image.

5. The method for detecting plastic particles based on image recognition according to claim 4, characterized in that: Construct a feedback verification model based on the estimated total weight of the image and the actual measured data of the weight sensor and perform calculations to obtain an artifact suspicion degree value, including the following steps: Process the composite effect of the image estimation error and the particle distribution fluctuation by introducing the hyperbolic tangent function and the logarithmic function to obtain a normalized response score, so as to enhance the response ability of the feedback verification model to the scenarios of uneven particle distribution or sudden change in image recognition deviation; Process the relative change rate of the actual measured weight in the current time period by introducing an exponential decay function to construct a feedback verification model, so as to suppress the misjudgment effect caused by the short-term fluctuation in the particle conveying process, and obtain an artifact suspicion degree value.

6. The method for detecting plastic particles based on image recognition according to claim 1, characterized in that: The determination status of the artifact interference includes no interference, acceptable interference, moderate artifact interference and severe artifact interference.

7. The method for detecting plastic particles based on image recognition according to claim 6, wherein: Determine the artifact interference in the image recognition model according to the estimated total weight of the image and the artifact suspicion degree value, including the following steps: Perform basic interference judgment according to the absolute error between the estimated total weight of the image and the actual measured data of the weight sensor. If the absolute error is less than or equal to the first error threshold W0 and the artifact suspicion degree value is less than or equal to the first interference threshold M0, then determine that the current state is no interference, and continue to perform the detection of plastic particles based on image recognition; If the absolute error is less than or equal to the first error threshold W0 and the artifact suspicion degree value is greater than the first interference threshold M0, then initially determine that the current state is acceptable interference and perform the next step; When the initial determination result is acceptable interference, perform interference escalation judgment based on the distribution uniformity coefficient of the image feature response value and the dispersion of the estimated weight of a single particle image. If the distribution uniformity coefficient is greater than or equal to the uniformity threshold A0 and the dispersion is less than or equal to the dispersion threshold B0, then determine that the current state is acceptable interference, and send a first-level prompt message to the operator, suggesting to perform a cleaning check on the image sensor; If the distribution uniformity coefficient is less than the uniformity threshold AO or the dispersion is greater than the dispersion threshold B0, then determine that the current state is moderate artifact interference, send a second-level warning message to the operator, and immediately trigger the parameter calibration process of the image recognition model; If the absolute error is greater than the first error threshold W0 and less than or equal to the second error threshold W0', and the artifact suspicion degree value is less than or equal to the second interference threshold M0', then initially determine that the current state is moderate artifact interference and perform the next step; When the initial determination result is moderate artifact interference, perform interference escalation judgment based on the proportion of the gray anomaly area in the image and the number of abnormal particles in the edge contour. If the proportion of the gray anomaly area is less than or equal to the area threshold C0 and the number of abnormal particles in the edge contour is less than or equal to the number threshold D0, then determine that the current state is moderate artifact interference, send a second-level warning message to the operator, and immediately trigger the parameter calibration process of the image recognition model; If the proportion of the gray anomaly area is greater than the area threshold C0 or the number of abnormal particles in the edge contour is greater than the number threshold D0, then determine that the current state is severe artifact interference, send a third-level alarm message, stop the current detection process and start the standby detection system; If the absolute error is greater than the second error threshold W0' or the artifact suspicion degree value is greater than the second interference threshold M0', then determine that the current state is severe artifact interference, send a third-level alarm message, stop the current detection process and start the standby detection system.

8. A plastic particle detection system based on image recognition, based on the plastic particle detection method according to any one of claims 1-7, characterized in that, Including: A data acquisition module, configured to acquire plastic particle image data and acquire the measured data of the particle weight sensor corresponding to the image data; An identification model construction module: configured to construct an image recognition model based on the image data to obtain the estimated total weight of the image; A calibration model construction module: configured to construct a feedback calibration model and perform calculations based on the estimated total weight of the image and the measured data of the weight sensor to obtain the artifact suspicion degree value; An intelligent detection module: configured to determine the artifact interference in the image recognition model according to the estimated total weight of the image and the artifact suspicion degree value, and implement the detection of plastic particles based on image recognition.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the method for detecting plastic particles based on image recognition according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the method for detecting plastic particles based on image recognition according to any one of claims 1-7.

Images