Linear edge sealing plate detection control system and method based on image recognition

Abstract

The invention relates to the technical field of linear edge sealing plate detection control, and discloses a linear edge sealing plate detection control system and method based on image recognition, and the system comprises a detection model database; an image acquisition module; an image recognition module; a data processing module; the data comparison module pays attention to the data determination module; a model optimization module; the detection process is combined with the attention degree of the user, and the attention content of the user on the linear edge sealing process is fully considered, for example, the improvement of the linear edge sealing process, the improvement of the linear edge sealing efficiency, the optimization of the linear edge sealing detection technology, the suitability of a selected detection model, the optimization of attention process parameters and the like. Therefore, the attention requirements of the user on different contents can be met.

Description

Technical Field

[0001] The present invention relates to the technical field of linear edge banding panel detection control, and in particular to a linear edge banding panel detection control system and method based on image recognition. Background Art

[0002] The linear edge banding process is a process of edge processing of materials by using laser or other technologies. By linear edge banding of the plate, it can be partially or completely melted and quickly cooled and solidified to form a smooth, uniform and burr-free straight edge.

[0003] The straight-line edge banding process can effectively provide high-quality edge effects, making the edges of the materials less susceptible to damage, more durable, and waterproof, moisture-proof and wear-resistant. In addition, the straight-line edge banding process can also achieve the advantage of not using glue or adhesives to seal the edges, avoiding environmental pollution and the generation of harmful gases.

[0004] In the prior art, the detection of the process parameters of the straight edge banding of the plate is often performed by taking a real-time image of the straight edge banding position of the plate through a visual camera, etc., and determining whether the process parameters of the straight edge banding position of the plate meet the requirements by comparing with the preset parameters, thereby determining whether the straight edge banding parameters of the plate are qualified; while such a detection scheme can quickly realize the detection process of the process parameters of the straight edge banding position of the plate, it still has the following defects and shortcomings: 1) The existing detection methods cannot reflect the user's concern. The existing detection methods can only quickly determine whether the process parameters of the linear edge banding position meet the preset requirements to determine whether it is qualified. However, the user's concern about the linear edge banding process may not be limited to whether the process parameters of the linear edge banding position are qualified. Their concern about the linear edge banding process may also include, for example, the improvement of the linear edge banding process, the improvement of the linear edge banding efficiency, and the optimization of the linear edge banding detection technology. The existing detection methods cannot reflect the user's concern and therefore cannot meet the user's above-mentioned concern needs; 2) The existing detection methods cannot effectively optimize the edge banding process that users are concerned about, nor can they optimize the detection process. The existing detection methods can only make the process parameters of the straight edge banding approach the preset threshold range, so that its detection results tend to be qualified, but cannot achieve real optimization for the user's concerns (such as whether the selected detection model is appropriate, whether the detection process can be optimized, whether the process parameters can be optimized, etc.); 3) Existing detection methods cannot provide users with detection methods and detection models that are adapted to their focus, making it difficult for the existing detection process to meet users' different focus content needs.

[0005] Therefore, it is urgent to provide a new solution to solve the defects and shortcomings in the above-mentioned prior art. Summary of the invention

[0006] In order to solve the defects and shortcomings in the prior art, the present invention provides a linear edge banding panel detection control system and method based on image recognition.

[0007] The technical solution provided by the present invention is as follows: a straight edge banding panel detection control system based on image recognition, characterized in that: the system comprises: a detection model database, wherein the detection model database stores a plurality of detection models for user selection, the detection model selected by the user determines the threshold range of data detection after training, and determines the preset data set of concern and the preset data set of non-concern according to the threshold range, and classifies and temporarily stores them; an image acquisition module, wherein the image recognition module performs image acquisition on the straight edge banding position of the panel and sends the acquired image to the image recognition module; an image recognition module, wherein the image recognition module obtains the required image data by sequentially performing image segmentation, image feature extraction, feature classification and recognition on the acquired image; a data processing module, wherein the data processing module processes the image The data is cleaned, screened, marked and summarized to distinguish valid data from invalid data, eliminate invalid data and retain valid data; a data comparison module, which sends the acquired valid data to the selected detection model and compares it with the determined threshold range, and determines the qualified detection data set and the unqualified detection data set according to the comparison result; a focus data determination module, which compares the qualified detection data set and the unqualified detection data set with the preset focus data set and the preset non-focus data set respectively, and adjusts the first parameter of the detection model selected by the user according to the comparison result; a model optimization module, which optimizes and corrects the second parameter of the detection model selected by the user according to the comparison result in the focus data determination module.

[0008] As a further preferred embodiment of the present invention, the detection model database includes: a model storage unit, in which a plurality of detection models for user selection are stored; a model selection unit, in which a plurality of detection models are sequentially displayed according to a recommendation index, and a user can select a detection model through the model selection unit; a model training unit, in which the model training unit trains the detection model selected by the user; a threshold determination unit, in which the threshold determination unit determines the threshold range of the trained detection model; a preset attention determination unit, in which the preset attention determination unit determines a preset data set requiring attention and a preset data set not requiring attention according to the threshold range of the detection model; and a classification temporary storage library, in which the classification temporary storage library classifies and temporarily stores data in the preset data set requiring attention and the preset data set not requiring attention.

[0009] As a further preferred embodiment of the present invention, in the preset attention determination unit, the preset attention-required data set and the preset non-attention data set are determined according to the following formula: in, Indicates the target user Data degree of attention; Indicates the most similar point to the user's attention relevant content; Indicates the user's focus on data the closest degree; Indicates user Contains target users the number of Indicates user Data The degree of attention of users Data Rating; and meet: When determining the data It belongs to the preset data set that needs attention; When determining the data It belongs to the preset non-attention data set; where Q is the attention judgment criterion related to the threshold range of the detection model.

[0010] As a further preferred embodiment of the present invention, in the focus data determination module, the first parameter includes at least a recommendation index based on which multiple detection models are sequentially displayed in the model selection unit; in the model optimization module, the second parameter includes at least a user's score for the data.

[0011] As a further preferred embodiment of the present invention, in the data of interest determination module, when all the qualified detection data sets or the unqualified detection data sets belong to the preset data sets that need attention, the first parameter of the detection model selected by the user is increased accordingly; when both the qualified detection data sets and the unqualified detection data sets do not belong to the preset data sets that need attention, the first parameter of the detection model selected by the user is decreased accordingly.

[0012] As a further preferred embodiment of the present invention, when all the qualified detection data sets or the unqualified detection data sets belong to the preset data sets that require attention, the single increase amplitude of the first parameter of the detection model selected by the user is greater than the single decrease amplitude of the first parameter of the detection model selected by the user when neither the qualified detection data sets nor the unqualified detection data sets belong to the preset data sets that require attention.

[0013] As a further preferred embodiment of the present invention, in the data of interest determination module, when a part of the qualified detection data set or the unqualified detection data set belongs to the preset data set requiring attention, and the other part belongs to the preset data set not requiring attention, and the proportion of data belonging to the preset data set requiring attention is higher than the proportion of data belonging to the preset data set not requiring attention, the model optimization module correspondingly increases the second parameter of the detection model selected by the user.

[0014] As a further preferred embodiment of the present invention, in the data of interest determination module, when a part of the qualified detection data set or the unqualified detection data set belongs to the preset data set requiring attention, and the other part belongs to the preset data set not requiring attention, and the proportion of data belonging to the preset data set requiring attention does not exceed the proportion of data belonging to the preset data set not requiring attention, the model optimization module correspondingly reduces the second parameter of the detection model selected by the user.

[0015] As a further preferred embodiment of the present invention, the model optimization module increases the amplitude of a single increase in the second parameter of the detection model selected by the user greater than the amplitude of a single decrease.

[0016] Furthermore, the present invention also provides a method for a linear edge banding panel detection control system based on image recognition, characterized in that it comprises the following steps: S100: The user selects a detection model from the detection model database, determines a threshold range for data detection after training the selected detection model, determines a preset data set of concern and a preset data set of no concern according to the threshold range, and classifies and temporarily stores them; S200: The image acquisition module acquires images of the linear edge banding position of the plate, and sends the acquired images to the image recognition module; S300: The image recognition module performs image segmentation, image feature extraction, feature classification and recognition on the collected image in sequence to obtain required image data; S400: The data processing module cleans, screens, marks and summarizes the image data to distinguish valid data from invalid data, removes invalid data and retains valid data; S500: The data comparison module sends the acquired valid data to the selected detection model and compares it with the determined threshold range, and determines the qualified detection data set and the unqualified detection data set according to the comparison result: S600: The attention data determination module compares the qualified detection data set and the unqualified detection data set with the preset attention-required data set and the preset non-attention data set respectively, and adjusts the first parameter of the detection model selected by the user according to the comparison result; S700: The model optimization module optimizes and modifies the second parameter of the detection model selected by the user according to the comparison result in the focus data determination module.

[0017] The beneficial effects achieved by the present invention compared with the prior art include: 1) The present invention provides a linear edge banding panel inspection control system and method based on image recognition. By providing a detection model database, it can determine a preset data set requiring attention and a preset data set not requiring attention according to a threshold range, and the set attention data determination module compares the detection qualified data set and the detection unqualified data set with the preset data set requiring attention and the preset data set not requiring attention respectively, and adjusts the first parameter of the detection model selected by the user according to the comparison result, thereby combining the detection process with the user's attention level, and fully considering the user's attention content on the linear edge banding process, such as the improvement of the linear edge banding process, the improvement of the linear edge banding efficiency, the optimization of the linear edge banding detection technology, whether the selected detection model is appropriate, whether the attention process parameters can be optimized, etc., so as to meet the user's attention needs on different contents; 2) The present invention provides a linear edge banding panel detection control system and method based on image recognition. By providing a detection model database, the first parameter of the detection model selected by the user can be adjusted according to the comparison result, so as to provide the user with a detection model adapted to his / her focus, so that the existing detection process is further adapted to the user's focus content requirements, thereby satisfying the user's focus content requirements; 3) The present invention provides a straight edge banding panel detection control system and method based on image recognition. By providing a model optimization module, the model optimization module can optimize and correct the second parameter of the detection model selected by the user according to the comparison result in the attention data determination module, so as to further optimize the user's attention level or score for the attention point according to the user's attention content, add the optimization content range introduced by the user's attention content to the original detection model or detection method, and also can further correct the user's attention level or score for the attention point. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work, among which: Figure 1 Shown is a schematic diagram of the structure of the system provided by the present invention; Figure 2 The figure shows a schematic diagram of the structure of the detection model database provided by the present invention; Figure 3 Shown is a flow chart of the steps of the method provided by the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "front end", "rear end", "two ends", "one end", "the other end" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0021] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "provided with", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0022] First embodiment: Figure 1 The first embodiment of the present invention provides a linear edge banding panel detection control system based on image recognition, the system includes: a detection model database, wherein the detection model database stores a plurality of detection models for user selection, the detection model selected by the user is trained to determine the threshold range of data detection, and the preset data set of concern and the preset data set of non-concern are determined according to the threshold range, and they are classified and temporarily stored; in this embodiment, Figure 2As shown, the detection model database includes: a model storage unit, in which a plurality of detection models for users to select are stored; each detection model includes detection parameters related to the straight edge banding of the panel, and the detection parameters may include, for example, three-dimensional parameters of the edge banding (length, width and thickness), shape parameters of the edge banding (edge ​​banding straightness and parallelism, etc.), area parameters of the edge banding (edge ​​banding area and edge banding area, etc.) and performance parameters of the edge banding (edge ​​banding smoothness and edge banding flatness, etc.); the detection models can be classified and stored in the model storage unit according to categories, or can be classified and stored according to other selected parameters, and other parameters can be selected such as frequency of use, update date and model size, etc., but the detection models need to be displayed in the model selection unit in sequence according to the recommended index; a model selection unit, in which a plurality of detection models are displayed in the model selection unit in sequence according to the recommended index, and the user can select the detection model through the model selection unit; by setting the corresponding recommended index for each detection model, Instead, a detection model adapted to the user's focus can be provided, so that the existing detection process can meet the user's different focus content requirements, and the user can successfully select a detection model adapted to the content requirements of his focus; a model training unit, the model training unit trains the detection model selected by the user; so that through the training process, the detection accuracy of the detection model selected by the user can meet the detection requirements of this time; a threshold determination unit, the threshold determination unit determines the threshold range of the trained detection model, and the subsequent data comparison module can compare the valid data with the determined threshold range through the determined threshold range, so as to determine the qualified detection data set and the unqualified detection data set according to the comparison result; a preset focus determination unit, the preset focus determination unit determines the preset focus data set and the preset non-focus data set according to the threshold range of the detection model; in the preset focus determination unit provided in this embodiment, the preset focus data set and the preset non-focus data set are determined according to the following formula: in, Indicates the target user Data degree of attention; Indicates the most similar point to the user's attention relevant content; Indicates the user's focus on data the closest degree; Indicates user Contains target users the number of Indicates user Data The degree of attention of users Data Rating; and meet: When determining the data It belongs to the preset data set that needs attention; When determining the data It belongs to the preset non-attention data set; where Q is the attention judgment criterion related to the threshold range of the detection model.

[0023] It is worth noting that, on the one hand, since the user's focus on the straight edge banding process may not be limited to whether the process parameters of the straight edge banding position are qualified, the user's focus on the straight edge banding process may also include, for example, the improvement of the straight edge banding process, the improvement of the straight edge banding efficiency, and the optimization of the straight edge banding detection technology. Therefore, the focus criterion Q is often not directly equivalent to the threshold range determined by the detection model. It can have multiple relationships with the threshold range determined by the detection model, such as intersection, inclusion, or non-intersection. For example, when the threshold range of the straight edge banding thickness determined by the detection model is [0.9mm, 1.1mm], the focus criterion Q can pay more attention to the influence of a certain range area within the threshold range. For example, at this time, the focus criterion Q can be set to be equal to the threshold range. The value range is [1.0mm, 1.1mm], so that the attention criterion Q at this time maintains an inclusion relationship with the threshold range determined by the detection model, and when the attention criterion Q is set to be related to the threshold range [0.7mm, 0.8mm], then at this time, there is no intersection between the attention criterion Q and the threshold range determined by the detection model; on the other hand, since Q is the attention criterion related to the threshold range of the detection model, it can be chosen to be related to the qualified threshold range or the unqualified threshold range, so the preset data set to be paid attention to can be related to the qualified detection data set or the unqualified detection data set; similarly, the preset non-attention data set can also be related to the qualified detection data set or the unqualified detection data set.

[0024] By determining the preset data sets that need attention and the preset data sets that are not of concern, the detection process is combined with the user's attention level, and the user's attention content on the linear edge banding process is fully considered, such as the improvement of the linear edge banding process, the improvement of the linear edge banding efficiency, the optimization of the linear edge banding detection technology, whether the selected detection model is appropriate, whether the process parameters can be optimized, etc., so as to meet the user's attention needs for different contents; a classified temporary storage library is used to classify and temporarily store the data in the preset data sets that need attention and the preset data sets that are not of concern, so as to facilitate the subsequent retrieval and calling of the data in the preset data sets that need attention and the preset data sets that are not of concern.

[0025] An image acquisition module, wherein the image recognition module acquires images of the straight edge-sealed positions of the panels and sends the acquired images to the image recognition module. Image acquisition can be completed by a visual camera disposed at the straight edge-sealed positions of the panels. To improve the accuracy and reliability of the acquired images, the visual cameras disposed at the straight edge-sealed positions of the panels can be provided with multiple groups in different directions and dimensions. Thus, by acquiring images from the straight edge-sealed positions of the panels in different directions and dimensions, the accuracy and reliability of the corresponding data parameters reflected in the acquired images can be further improved. An image recognition module, wherein the image recognition module performs image segmentation on the acquired images in sequence. , image feature extraction and feature classification and recognition, to obtain the required image data; data processing module, the data processing module cleans, screens, marks and summarizes the image data to distinguish valid data from invalid data, removes invalid data and retains valid data; data comparison module, the data comparison module sends the acquired valid data to the selected detection model and compares it with the determined threshold range, and determines the qualified detection data set and the unqualified detection data set according to the comparison result; attention data determination module, the attention data determination module compares the qualified detection data set and the unqualified detection data set with the preset attention data set respectively The detection model database is provided so as to adjust the first parameter of the detection model selected by the user according to the comparison result, thereby providing the user with a detection model adapted to the user's focus point, so that the existing detection process is further adapted to the user's focus content requirements, thereby satisfying the user's focus content requirements; a model optimization module, wherein the model optimization module optimizes and modifies the second parameter of the detection model selected by the user according to the comparison result in the focus data determination module. The second parameter in the embodiment at least includes the user's score for the data. The model optimization module is provided so as to optimize and modify the second parameter of the detection model selected by the user according to the comparison result in the focus data determination module, thereby further optimizing the user's focus degree or score for the focus point according to the user's focus content, adding the optimization content range introduced by the user's focus content to the original detection model or detection method, and further modifying the user's focus degree or score for the focus point.

[0026] For example, when the threshold range determined by the threshold determination unit is [A, B], B>A, that is, when the detection parameter S satisfies A≤S≤B, the detection parameter is sent to the qualified detection data set, otherwise the detection parameter is sent to the unqualified detection data set, and due to the intervention of the user's attention level, the user's attention range may be [AP, B+P], and AP<B+P is satisfied. The optimized content range introduced by this part due to the user's attention content may be a positive optimized content range caused by the threshold range not available in the detection model selected due to the user's understanding of the industry, or it may be a negative content range that may be caused by improper setting of the user's attention level or score to the focus point. Therefore, the second parameter of the detection model is optimized and corrected, so that the user's attention level or score to the focus point can be further optimized according to the user's attention content, and the optimized content range introduced by the user's attention content is added to the original detection model or detection method, and the user's attention level or score to the focus point can also be further corrected.

[0027] In the focus data determination module provided by the present embodiment, when all the qualified detection data sets or the unqualified detection data sets belong to the preset data sets that need to be paid attention to, the first parameter of the detection model selected by the user is increased accordingly; thereby, the recommendation index of the detection model can be increased accordingly. When the user selects the detection model, a higher recommendation index can be more convenient for the user to select, so that the detection model selected by the user can be more adapted to his / her focus, so that the detection process performed by the detection model can further adapt to the user's focus content requirements, thereby meeting the user's focus content requirements; when neither the qualified detection data set nor the unqualified detection data set belongs to the preset data set that needs to be paid attention to, the first parameter of the detection model selected by the user is lowered accordingly, thereby, the recommendation index of the detection model can be lowered accordingly. When the user selects the detection model, a lower recommendation index is more unfavorable for the detection model to be selected by the user, so that the user can select a detection model that is more adapted to his / her focus, so as to further meet the user's focus content requirements.

[0028] As a further preferred implementation manner of this embodiment, when all the qualified data sets or the unqualified data sets belong to the preset data sets that require attention, the single increase amplitude of the first parameter of the detection model selected by the user is greater than the single decrease amplitude of the first parameter of the detection model selected by the user when neither the qualified data set nor the unqualified data set belongs to the preset data sets that require attention. This is set because, on the one hand, when neither the qualified data set nor the unqualified data set belongs to the preset data sets that require attention, this may be due to the data mismatch caused by the user's incorrect selection of the focus content or the focus content and the detection data not belonging to the same category (for example, the detection data is the edge sealing thickness, and the user's focus content is the smoothness of the sealing board, etc.), which should not affect the recommendation index of the detection model; on the other hand, there may be judgment criteria or threshold ranges that are not set in the detection model, and the user may reasonably supplement the judgment criteria or threshold ranges that have not been set in the detection model due to his knowledge or understanding of the industry. Therefore, at this time, it should not affect the equal amplitude reduction of the recommendation index of the detection model.

[0029] In the focus data determination module, when a part of the qualified detection data set or the unqualified detection data set belongs to the preset data set that needs attention, and the other part belongs to the preset non-focus data set, and the proportion of data belonging to the preset data set that needs attention is higher than the proportion of data belonging to the preset non-focus data set, the model optimization module correspondingly increases the second parameter of the detection model selected by the user. At this time, it means that most of the data in the qualified detection data set or the unqualified detection data set belongs to the preset data set that needs attention, which means that the user's score for the focus data is relatively reliable. Therefore, the score can be further improved, so that the user's attention level or score for the focus point can be further optimized according to the user's focus content.

[0030] In the focus data determination module, when a part of the qualified detection data set or the unqualified detection data set belongs to the preset data set that needs attention, and the other part belongs to the preset non-focus data set, and the proportion of data belonging to the preset data set that needs attention does not exceed the proportion of data belonging to the preset non-focus data set, the model optimization module correspondingly reduces the second parameter of the detection model selected by the user. At this time, it means that most of the data in the qualified detection data set or the unqualified detection data set belongs to the preset non-focus data set, which means that the user's score for the focus data is relatively unreliable. Therefore, it is necessary to further reduce the score so that the user's attention level or score for the focus point can be further optimized according to the user's correct focus content.

[0031] As a further preference of the present embodiment, the model optimization module sets a single increase amplitude greater than a single decrease amplitude for the second parameter of the detection model selected by the user; similarly, this is because, on the one hand, this may be due to the fact that the user's focus content and the detection data do not belong to the same category, resulting in data mismatch (for example, the detection data is edge sealing thickness, while the user's focus content is edge sealing smoothness, etc.), which should not affect the user's score for the focus content; on the other hand, there may be judgment criteria or threshold ranges that are not set in the detection model, and the user has made reasonable supplements to the judgment criteria or threshold ranges that have not been set in the detection model due to his knowledge or understanding of the industry. Therefore, this should not affect the user's score for the focus content. Therefore, it is necessary to further set the model optimization module to set a single increase amplitude greater than a single decrease amplitude for the second parameter of the detection model selected by the user.

[0032] Second embodiment: Figure 3 As shown, the second embodiment of the present invention further provides a method for the linear edge banding panel detection and control system based on image recognition mentioned in the first embodiment, comprising the following steps: S100: The user selects a detection model from the detection model database, determines a threshold range for data detection after training the selected detection model, determines a preset data set of concern and a preset data set of no concern according to the threshold range, and classifies and temporarily stores them; S200: The image acquisition module acquires images of the linear edge banding position of the plate, and sends the acquired images to the image recognition module; S300: The image recognition module performs image segmentation, image feature extraction, feature classification and recognition on the collected image in sequence to obtain required image data; S400: The data processing module cleans, screens, marks and summarizes the image data to distinguish valid data from invalid data, removes invalid data and retains valid data; S500: The data comparison module sends the acquired valid data to the selected detection model and compares it with the determined threshold range, and determines the qualified detection data set and the unqualified detection data set according to the comparison result: S600: The attention data determination module compares the qualified detection data set and the unqualified detection data set with the preset attention-required data set and the preset non-attention data set respectively, and adjusts the first parameter of the detection model selected by the user according to the comparison result; S700: The model optimization module optimizes and modifies the second parameter of the detection model selected by the user according to the comparison result in the focus data determination module.

[0033] It should be noted that the "one embodiment" or "embodiment" mentioned in the specification means that the specific features, structures or characteristics described in conjunction with the embodiment are included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment" or "embodiment" appearing in various places throughout the specification do not necessarily refer to the same embodiment.

[0034] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that these specific embodiments are only illustrative, and those skilled in the art may omit, replace, and change the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, merging the above method steps so as to perform substantially the same functions in substantially the same manner to achieve substantially the same results is within the scope of the present invention. Therefore, the scope of the present invention is limited only by the appended claims.

Claims

1. The linear edge banding panel detection and control system based on image recognition is characterized by: The system comprises: a detection model database, wherein a plurality of detection models for user selection are stored in the detection model database, the detection model selected by the user is trained to determine a threshold range for data detection, and a preset data set requiring attention and a preset data set not requiring attention are determined according to the threshold range, and the data sets are classified and temporarily stored; An image acquisition module, wherein the image acquisition module acquires images of the straight edge banding position of the plate and sends the acquired images to an image recognition module; an image recognition module, wherein the image recognition module acquires required image data by sequentially performing image segmentation, image feature extraction, feature classification and recognition on the acquired images; a data processing module, wherein the data processing module cleans, screens, marks and summarizes the image data to distinguish valid data from invalid data, removes invalid data and retains valid data; a data comparison module, wherein the data comparison module sends the acquired valid data to a selected detection model and compares it with a determined threshold range, and determines a qualified detection data set and an unqualified detection data set according to the comparison result; a data determination module for attention, wherein the data determination module for attention adjusts a first parameter of the detection model selected by the user according to the comparison result by comparing the qualified detection data set and the unqualified detection data set with a preset data set to be paid attention to and a preset data set not to be paid attention to; and a model optimization module, wherein the model optimization module optimizes and corrects a second parameter of the detection model selected by the user according to the comparison result in the data determination module for attention.

2. The linear edge banding panel detection and control system based on image recognition according to claim 1 is characterized in that: The detection model database includes: a model storage unit, in which a plurality of detection models for user selection are stored; a model selection unit, in which a plurality of detection models are sequentially displayed according to a recommendation index, and a user can select a detection model through the model selection unit; a model training unit, in which the model training unit trains the detection model selected by the user; a threshold determination unit, in which the threshold determination unit determines the threshold range of the trained detection model; a preset attention determination unit, in which the preset attention determination unit determines a preset data set requiring attention and a preset data set not requiring attention according to the threshold range of the detection model; and a classification temporary storage library, in which the classification temporary storage library classifies and temporarily stores data in the preset data set requiring attention and the preset data set not requiring attention.

3. The linear edge banding panel detection and control system based on image recognition according to claim 2 is characterized in that: In the preset attention determination unit, the preset attention-required data set and the preset non-attention data set are determined according to the following formula: in, Indicates the target user Data degree of attention; Indicates the most similar point to the user's attention relevant content; Indicates the user's focus on data the closest degree; Indicates user Contains target users the number of Indicates user Data The degree of attention of users Data Rating; and meet: When determining the data It belongs to the preset data set that needs attention; When determining the data It belongs to the preset non-attention data set; where Q is the attention judgment criterion related to the threshold range of the detection model.

4. The linear edge banding panel detection and control system based on image recognition according to claim 3 is characterized in that: In the focus data determination module, the first parameter includes at least a recommendation index based on which multiple detection models are sequentially displayed in the model selection unit; in the model optimization module, the second parameter includes at least a user's rating of the data.

5. The linear edge banding panel detection and control system based on image recognition according to claim 4 is characterized in that: In the concerned data determination module, when all the qualified data sets or unqualified data sets belong to the preset concerned data sets, the first parameter of the detection model selected by the user is increased accordingly; When neither the qualified data set nor the unqualified data set belongs to the preset data set requiring attention, the first parameter of the detection model selected by the user is correspondingly reduced.

6. The linear edge banding panel detection and control system based on image recognition according to claim 5 is characterized in that: When all the qualified data sets or the unqualified data sets belong to the preset data sets that require attention, the single increase amplitude of the first parameter of the detection model selected by the user is greater than the single decrease amplitude of the first parameter of the detection model selected by the user when neither the qualified data sets nor the unqualified data sets belong to the preset data sets that require attention.

7. The linear edge banding panel detection and control system based on image recognition according to claim 4 is characterized in that: In the data of interest determination module, when a part of the qualified data set or the unqualified data set belongs to the preset data set that needs attention, and the other part belongs to the preset data set that does not need attention, and the proportion of data belonging to the preset data set that needs attention is higher than the proportion of data belonging to the preset data set that does not need attention, the model optimization module correspondingly increases the second parameter of the detection model selected by the user.

8. The linear edge banding panel detection and control system based on image recognition according to claim 7 is characterized in that: In the data of interest determination module, when a part of the qualified data set or the unqualified data set belongs to the preset data set that requires attention, and the other part belongs to the preset data set that is not of interest, and the proportion of data belonging to the preset data set that requires attention does not exceed the proportion of data belonging to the preset data set that is not of interest, the model optimization module correspondingly reduces the second parameter of the detection model selected by the user.

9. The linear edge banding panel detection and control system based on image recognition according to claim 8, characterized in that: The model optimization module increases the amplitude of a single increase in the second parameter of the detection model selected by the user greater than the amplitude of a single decrease.

10. A method for a linear edge banding panel detection and control system based on image recognition as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: S100: The user selects a detection model from the detection model database, determines a threshold range for data detection after training the selected detection model, determines a preset data set of concern and a preset data set of no concern according to the threshold range, and classifies and temporarily stores them; S200: The image acquisition module acquires images of the linear edge banding position of the plate, and sends the acquired images to the image recognition module; S300: The image recognition module performs image segmentation, image feature extraction, feature classification and recognition on the collected image in sequence to obtain required image data; S400: The data processing module cleans, screens, marks and summarizes the image data to distinguish valid data from invalid data, removes invalid data and retains valid data; S500: The data comparison module sends the acquired valid data to the selected detection model and compares it with the determined threshold range, and determines the qualified detection data set and the unqualified detection data set according to the comparison result: S600: The attention data determination module compares the qualified detection data set and the unqualified detection data set with the preset attention-required data set and the preset non-attention data set respectively, and adjusts the first parameter of the detection model selected by the user according to the comparison result; S700: The model optimization module optimizes and modifies the second parameter of the detection model selected by the user according to the comparison result in the focus data determination module.

Images