Metal furniture quality inspection visualization system based on artificial intelligence

Abstract

The invention discloses a metal furniture quality inspection visualization system based on artificial intelligence, which belongs to the technical field of metal furniture quality inspection, and comprises a characteristic analysis module, a light source control module, an image acquisition module, a detection and analysis module and a display module; the characteristic analysis module is used for performing characteristic analysis on various metal accessories of metal furniture, determining the metal characteristics of various metal accessories, and adjusting the light source equipment based on the metal characteristics of various metal accessories; the light source control module is used for performing light source control analysis; the image acquisition module is used for performing image acquisition on metal accessories to obtain accessory acquisition images; the detection and analysis module is used for performing detection and analysis, segmenting the accessory acquisition images according to detection areas, and obtaining detection and analysis images; analyzing the detection and analysis images, and obtaining contour detection results and appearance detection results; the display module is used for performing data display, acquiring the acquisition data of various display items in real time, and displaying the various acquisition data in a display mode.

Description

Technical Field

[0001] The invention belongs to the technical field of metal furniture quality detection, and in particular is a metal furniture quality detection visualization system based on artificial intelligence. Background Art

[0002] With the rapid development of the manufacturing industry and the increasing demand of consumers for product quality, metal furniture, as an important part of the furniture industry, has become particularly important in terms of quality inspection. However, the traditional quality inspection method for metal furniture mainly relies on manual visual inspection and simple measuring tools. Manual visual inspection takes a lot of time and energy, and the inspection speed is limited by personnel fatigue and experience, which makes it difficult to meet the needs of large-scale production lines. Moreover, the inspection results are greatly affected by the subjective judgment of the inspectors, which is prone to misjudgment and missed judgment, affecting the stability and reliability of product quality. In order to improve the accuracy of inspection, companies often need to hire a large number of experienced inspectors and equip them with expensive inspection equipment, resulting in high inspection costs. At the same time, traditional inspection methods can only provide simple text or digital descriptions, lacking intuitive visual display, which is not conducive to the rapid location and resolution of problems.

[0003] Based on this, in order to solve the above problems, the present invention provides a metal furniture quality inspection visualization system based on artificial intelligence. Summary of the invention

[0004] In order to solve the problems existing in the above solutions, the present invention provides a metal furniture quality inspection visualization system based on artificial intelligence.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] The metal furniture quality inspection visualization system based on artificial intelligence includes a characteristic analysis module, a light source control module, an image acquisition module, a detection and analysis module, and a display module;

[0007] The characteristic analysis module is used to perform characteristic analysis on each metal accessory of the metal furniture, determine the metal characteristic of each metal accessory, and adjust the light source device based on the metal characteristic of each metal accessory.

[0008] The light source control module is used to perform light source control analysis, identify the metal characteristics of the metal accessories to be inspected, match the corresponding light source adjustment method according to the metal characteristics; adjust the light source according to the light source adjustment method; perform real-time light source calibration during the inspection of metal accessories to obtain corresponding calibration results, including light control qualified and light control unqualified;

[0009] When the calibration result shows that the light control is qualified, no corresponding operation is performed;

[0010] When the verification result shows that the light control is unqualified, the light control is adjusted and a corresponding light control verification notification is sent to the image acquisition module.

[0011] Furthermore, the method for performing real-time light source calibration during the detection of metal accessories includes:

[0012] Identify the metal accessory information to be detected, and adjust the detection area according to the metal accessory information; set the light control accessories in the detection area according to the metal accessory information;

[0013] During the inspection of metal accessories, real-time image acquisition is performed on the light-control accessories to obtain light-control analysis images of the light-control accessories;

[0014] A preset light control standard image is obtained, and a real-time calibration analysis is performed on the collected light control analysis image based on the light control standard image to obtain a corresponding calibration result.

[0015] Furthermore, the method for adjusting the detection area according to the metal accessory information includes:

[0016] According to the information of each metal accessory, a corresponding simulation part is set, and the image of the simulation part is captured under preset conventional lighting conditions and re-inspection lighting conditions to obtain a conventional capture simulation image and a re-inspection capture simulation image;

[0017] Perform image recognition on the conventional acquisition simulation image and the re-inspection acquisition simulation image respectively, determine the area where the image recognition is abnormal, and mark it as the abnormal simulation area; determine the conventional simulation range and the re-inspection simulation range according to each abnormal simulation area in the conventional acquisition simulation image and the re-inspection acquisition simulation image respectively, and set the initial detection area according to the conventional simulation range and the re-inspection simulation range;

[0018] The size information of the metal accessory information is identified, and the initial detection area is adjusted according to the size information to obtain the detection area.

[0019] Furthermore, the method for setting the light control accessory includes:

[0020] The reference parts are set according to the information of each metal accessory; the part information of each reference part is identified, and the part information of each reference part is summarized to set the light control accessory; and the light control accessory is installed in the detection area.

[0021] Furthermore, the method for performing real-time verification and analysis on the collected light control analysis image based on the light control standard image includes:

[0022] The light control calibration model is established based on the light control standard image. The expression of the light control standard model is:

[0023] ;

[0024] Where: s is the input data, the input data is the light control analysis image; the output data is the light control calibration value GK(s);

[0025] The light control analysis image is analyzed in real time by using a light control calibration model to obtain a corresponding light control calibration value, and a calibration result is determined according to the light control calibration value.

[0026] The image acquisition module is used to acquire images of metal accessories, acquire images of metal accessories within the detection area, and obtain acquired accessory images; and send the acquired accessory images to the detection and analysis module.

[0027] Furthermore, when the image acquisition module receives the light control verification notification sent by the light source control module, it identifies the corresponding accessory acquisition image, marks it as an abnormal accessory acquisition image, removes the abnormal accessory acquisition image, and re-acquires the corresponding accessory acquisition image.

[0028] The detection and analysis module is used to perform detection and analysis, obtain the accessory acquisition image, segment the accessory acquisition image according to the detection area, and obtain the detection and analysis image; perform grayscale processing on the detection and analysis image to obtain a grayscale detection image;

[0029] Identify the boundary contour of the metal accessories and the positioning boundary of the light control accessories according to the grayscale detection image; obtain the size information of the light control accessories, and generate a standard contour based on any inflection point of the boundary contour according to the positioning boundary, light control size and design size; check the boundary contour according to the standard contour to obtain the contour detection result; the contour detection result includes unqualified contour and qualified contour;

[0030] When the contour detection result is that the contour is unqualified, no subsequent analysis is performed;

[0031] When the contour detection result is that the contour is qualified, the appearance inspection of the metal accessories is performed according to the grayscale detection image to obtain the corresponding appearance inspection result.

[0032] Furthermore, the method for identifying the boundary contour of the metal fitting according to the grayscale detection image includes:

[0033] Identify the grayscale value corresponding to each pixel in the grayscale detection image, calculate the grayscale difference between each pixel and each adjacent pixel, and integrate the grayscale differences between each adjacent pixel into a positioning set of the corresponding pixel;

[0034] Set boundary judgment requirements and establish a boundary judgment model based on the boundary judgment requirements. The expression of the boundary judgment model is: ;

[0035] Where: q is the input data, the input data is the positioning set; the output data is the boundary judgment value BP(q);

[0036] The positioning set of each pixel is analyzed through the boundary judgment model to obtain the boundary judgment value of each pixel; the boundary contour of the metal accessory is determined according to the boundary judgment value of each pixel.

[0037] Furthermore, the method for performing appearance inspection on metal accessories according to the grayscale inspection image includes:

[0038] Identify each color block in the metal accessories, segment the grayscale detection image according to the boundary contour and each color block, and obtain the grayscale area image corresponding to each color block;

[0039] Identify the grayscale value of each pixel in the grayscale area image; calculate the location set of each pixel, perform pixel evaluation based on each location set to obtain each abnormal pixel; determine the appearance evaluation result based on the distribution of each abnormal pixel.

[0040] The display module is used for displaying data, setting various display items and corresponding display modes, acquiring collected data of various display items in real time, and displaying various collected data according to the display mode.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] Through the cooperation among the characteristic analysis module, light source control module, image acquisition module, detection and analysis module and display module, intelligent quality detection of metal furniture can be realized, and the detection results can be displayed to users more intuitively. By combining the light control boundary to generate a standard contour, the contour of the metal accessories is detected based on the standard contour, which will realize multiple detections such as size and deformation, and the standard contour and boundary contour can be displayed to the user more intuitively; rather than just one size comparison result. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] 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 will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 This is a principle block diagram of the present invention. DETAILED DESCRIPTION

[0045] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. 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.

[0046] like Figure 1 As shown, the metal furniture quality inspection visualization system based on artificial intelligence includes a characteristic analysis module, a light source control module, an image acquisition module, a detection and analysis module and a display module;

[0047] The characteristic analysis module is used to perform characteristic analysis on each metal accessory of metal furniture to understand the relevant characteristics of the metal accessories during the production and testing process; obtain information on each metal accessory, including material, color, size, connection method and other related information; determine various quality inspection problems that may exist according to the metal accessory information, such as scratches, size discrepancies, dents and other quality inspection problems, obtain the collected images corresponding to each quality inspection problem, and determine whether it will be unable to be identified by the current conventional detection due to characteristics such as color, such as minor scratches, stains with small color differences, etc., which may be difficult to identify due to small differences in color, problem contrast, etc., and divide each quality inspection problem into a first category and a second category according to whether it can be identified. The first category is conventional quality inspection problems, and the second category is various quality inspection problems that may not be identified due to characteristics such as color. classification; according to the recognition of each quality inspection problem in the second classification under other colors, angles and other different lighting conditions, determine various ways to highlight the characteristics of the problem, summarize and count based on various methods, and determine the corresponding re-inspection light source adjustment method, that is, what color of light source, angle, etc. to use for illumination re-inspection; generally, professionals analyze and set the re-inspection light source adjustment method according to the actual nature of various quality inspection problems, and try to add an auxiliary light source device to achieve the above-mentioned re-inspection light source adjustment method; it needs to be determined in combination with the detection conditions; according to the first classification, the second classification and each re-inspection light source adjustment method, the metal characteristics of each metal accessory are set. The metal characteristics are generally used to indicate which quality inspection problems need to be re-inspected and the corresponding re-inspection methods; the light source equipment is adjusted and supplemented according to the metal characteristics of each metal accessory.

[0048] In one embodiment, the metal properties of each metal accessory may be directly set based on existing technologies, conclusions, etc., and the light source device may be adjusted and supplemented at the same time.

[0049] The light source control module is used to perform light source control analysis, identify the metal accessories information to be detected, and adjust the detection area according to the metal accessories information. The detection area is the area where the metal accessories are collected and detected;

[0050] According to the information of each metal accessory, a light control accessory is set in the detection area. The light control accessory is a physical reference structure set according to the material, color, etc. in the metal accessory information;

[0051] Identify the metal characteristics of the metal accessories to be inspected, and match the corresponding light source adjustment method according to the metal characteristics. For example, first apply a conventional light source, and then adjust it according to the corresponding re-inspection light source adjustment method. Specific adjustments are made according to actual conditions; adjust the light source according to the light source adjustment method;

[0052] During the inspection of metal accessories, real-time image acquisition is performed on the light control accessories to obtain a real-time acquired image of the light control accessories, which is marked as a light control analysis image;

[0053] Acquire a preset light control standard image, that is, a captured image of the light control accessory under normal conditions; perform real-time calibration and analysis on the captured light control analysis image based on the light control standard image to obtain corresponding calibration results, including light control qualified and light control unqualified; perform direct calibration judgment based on image comparison; the judgment can be made based on existing image recognition technology; illustratively, establish a light control calibration model based on the light control standard image, which is used to calibrate the light control analysis image based on the light control standard image, and if it meets the calibration standard, it is considered to have passed the calibration and is considered to be light control qualified; the expression of the light control standard model is: ; Where: s is the input data, the input data is the light control analysis image; the output data is the light control calibration value GK(s); the calibration result is determined according to the light control calibration value;

[0054] When the calibration result shows that the light control is qualified, no corresponding operation is performed;

[0055] When the verification result shows that the light control fails, the light control is adjusted; it can be a manual adjustment or an intelligent adjustment based on existing artificial intelligence to make the light control qualified; such as pre-analyzing the adjustment method corresponding to each light control problem, and subsequently performing adjustment evaluation; and sending a corresponding light control verification notification to the image acquisition module, the light control verification notification includes the time-related data corresponding to the failure.

[0056] Methods for adjusting the detection area according to metal fittings information include:

[0057] Preparation before testing;

[0058] Perform acquisition simulation according to the metal accessories information, that is, set a large-area simulation part that meets the inspection standards according to the material, color, etc. in the metal accessories information, and perform image acquisition on the simulation part under preset conventional lighting conditions and re-inspection lighting conditions to obtain conventional acquisition simulation images and re-inspection acquisition simulation images;

[0059] Identify the conventionally collected simulated images and the re-inspection collected simulated images respectively, and determine the areas with image recognition anomalies, such as reflections, shadows, etc., which are caused by the relationship between position and light source; use existing image recognition technology to determine; and because the number of simulations in the early stage is not large, it can also be directly marked manually;

[0060] Mark the area with abnormal image recognition as the abnormal simulation area; determine the optimal acquisition range corresponding to the normal lighting condition and the re-inspection lighting condition respectively according to the abnormal simulation area, that is, image recognition abnormality may occur if the optimal acquisition range is exceeded, and mark the corresponding optimal acquisition range as the normal simulation range and the re-inspection simulation range respectively, and take the intersection area between the normal simulation range and the re-inspection simulation range as the initial detection area;

[0061] Adjustments during testing;

[0062] Identify the size information of metal accessories, adjust the initial detection area according to the size information, and obtain the detection area; that is, reduce the initial detection area to reduce the subsequent data processing volume, and only process the image data within the detection area.

[0063] The method of setting the light control accessories in the detection area according to the metal accessories information includes:

[0064] According to historical image acquisition data and structural features, determine which parts of the metal accessories, such as structure and color, are susceptible to light, thereby affecting the detection accuracy, and mark the corresponding parts as reference parts; identify the part information of each reference part, and set the light control accessories according to the part information of each reference part; that is, integrate each reference part into the same structure and mark it as a light control accessory; and the light control accessory has a preset standard size to balance the scale problem of the image and improve the size recognition accuracy;

[0065] Install the light control accessories in the detection area where they will not affect the detection of metal accessories.

[0066] The image acquisition module is used to acquire images of metal accessories, acquire images of metal accessories within the detection area, and obtain acquired images of accessories; and send the acquired images of accessories to the detection and analysis module.

[0067] In one embodiment, when the image acquisition module receives the light control verification notification sent by the light source control module, it identifies the corresponding accessory acquisition image, marks it as an abnormal accessory acquisition image, removes the abnormal accessory acquisition image, and re-acquires the corresponding accessory acquisition image.

[0068] The detection and analysis module is used to perform detection and analysis, obtain the accessory acquisition image, segment the accessory acquisition image according to the detection area, and obtain the detection and analysis image; because the light control accessory is set in the detection area, the light control accessory in the detection and analysis image can be identified; because the background color of the detection area and the color of the metal accessory have a large difference in grayscale expression, they have a large grayscale value difference; grayscale processing is performed on the detection and analysis image to obtain a grayscale detection image;

[0069] According to the grayscale detection image, the boundary contour of the metal accessory and the positioning boundary of the light control accessory are identified; the light control size of the light control accessory and the design size of the metal accessory are obtained, and a standard contour is generated according to any inflection point of the boundary contour according to the positioning boundary, the light control size and the design size; because the corresponding scale, inclination and other related parameters can be determined according to the image length and actual size of the positioning boundary, the corresponding inflection point is used as the reference point, and the corresponding design boundary is generated according to the angle and design size of the boundary contour, and finally a standard contour corresponding to the boundary contour is formed; the boundary contour is checked according to the standard contour to obtain the contour detection result, and the contour detection result includes unqualified contour and qualified contour; by combining the light control boundary to generate the standard contour, and detecting the contour of the metal accessory based on the standard contour, multiple detections such as size and deformation will be realized, and the standard contour and boundary contour can be more intuitively displayed to the user; rather than just one size comparison result.

[0070] When the contour detection result is that the contour is unqualified, no subsequent analysis is performed;

[0071] When the contour detection result is that the contour is qualified, the appearance inspection of the metal accessories is performed according to the grayscale detection image to obtain the corresponding appearance inspection result.

[0072] The method of identifying the boundary contour of metal accessories based on the grayscale detection image includes:

[0073] Identify the grayscale value corresponding to each pixel in the grayscale detection image, calculate the grayscale difference between each pixel and its adjacent pixels, and integrate the grayscale difference between each adjacent pixel into the positioning set of the pixel. If the grayscale difference is 1, 2, and 3 respectively, the positioning set is (1, 2, 3). Because the grayscale difference of the boundary contour will have a large difference, the boundary judgment requirements can be set according to its characteristics, and the boundary judgment model is established according to the boundary judgment requirements. The expression of the boundary judgment model is: ; Where: q is the input data, the input data is the positioning set; the output data is the boundary judgment value BP(q);

[0074] The positioning set of each pixel is analyzed through the boundary judgment model to obtain the boundary judgment value of each pixel; the boundary contour of the metal accessory is determined according to the boundary judgment value of each pixel.

[0075] The positioning boundary of the light control accessories is the same as the boundary contour recognition method of the metal accessories, so no corresponding description is given.

[0076] In one embodiment, the method for performing appearance inspection on metal accessories based on grayscale inspection images is: performing inspection and analysis based on existing image appearance inspection technology.

[0077] In one embodiment, a method for performing appearance inspection on a metal accessory according to a grayscale inspection image includes:

[0078] Identify the color blocks in the metal accessories, which refer to the area blocks of the same color; segment the grayscale detection image according to the boundary contour and each color block, and obtain the grayscale area image corresponding to each color block. Because the contour check is passed, the boundary of each color block can be located according to each color block and the boundary contour, and then the boundary is adjusted according to a certain grayscale difference, and then segmented to form each grayscale area image;

[0079] Identify the grayscale value of each pixel in the grayscale area image; calculate the positioning set of each pixel, and evaluate the pixel according to each positioning set to determine whether it is an abnormal situation. Therefore, the grayscale difference of the same color is very small. When the difference is too large, the pixel can be determined to be an abnormal pixel; obtain each abnormal pixel; determine the appearance evaluation result according to the distribution of each abnormal pixel; that is, determine whether it exceeds the appearance detection standard according to the size and length of the corresponding area, and then determine the appearance evaluation result.

[0080] The display module is used for data display, setting each display item and the corresponding display mode according to the user's display requirements, acquiring the collected data of each display item in real time, and displaying each collected data according to a preset display mode.

[0081] Through the cooperation among the characteristic analysis module, light source control module, image acquisition module, detection and analysis module and display module, intelligent quality detection of metal furniture can be realized, and the detection results can be displayed to users more intuitively.

[0082] The above formulas are all calculated by removing dimensions and taking numerical values. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data.

[0083] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. The metal furniture quality inspection visualization system based on artificial intelligence is characterized by: It includes a characteristic analysis module, a light source control module, an image acquisition module, a detection and analysis module and a display module; The characteristic analysis module is used to perform characteristic analysis on each metal accessory of the metal furniture, determine the metal characteristics of each metal accessory, and adjust the light source device based on the metal characteristics of each metal accessory; The light source control module is used to perform light source control analysis, identify the metal characteristics of the metal accessories to be inspected, match the corresponding light source adjustment method according to the metal characteristics; adjust the light source according to the light source adjustment method; perform real-time light source calibration during the inspection of metal accessories to obtain corresponding calibration results, including light control qualified and light control unqualified; When the calibration result shows that the light control is qualified, no corresponding operation is performed; When the verification result shows that the light control is unqualified, the light control is adjusted and a corresponding light control verification notification is sent to the image acquisition module; The image acquisition module is used to acquire images of metal accessories, acquire images of metal accessories within the detection area, and obtain acquired images of accessories; Sending the accessory captured image to a detection and analysis module; The detection and analysis module is used to perform detection and analysis, obtain the accessory acquisition image, segment the accessory acquisition image according to the detection area, and obtain the detection and analysis image; Performing grayscale processing on the detection and analysis image to obtain a grayscale detection image; Identify the boundary contour of the metal accessories and the positioning boundary of the light control accessories according to the grayscale detection image; obtain the size information of the light control accessories, and generate a standard contour based on any inflection point of the boundary contour according to the positioning boundary, light control size and design size; check the boundary contour according to the standard contour to obtain the contour detection result; the contour detection result includes unqualified contour and qualified contour; When the contour detection result is that the contour is unqualified, no subsequent analysis is performed; When the contour detection result is qualified, the metal accessories are inspected for appearance according to the grayscale detection image to obtain the corresponding appearance inspection result; The display module is used to display data, set various display items and corresponding display modes, obtain collected data of various display items in real time, and display various collected data according to the display mode; Methods for real-time light source calibration during metal parts inspection include: Identify the metal accessory information to be detected, and adjust the detection area according to the metal accessory information; set the light control accessories in the detection area according to the metal accessory information, and the light control accessories are a physical reference structure set according to the material and color in the metal accessory information; During the inspection of metal accessories, real-time image acquisition is performed on the light-control accessories to obtain light-control analysis images of the light-control accessories; Obtain a preset light control standard image, perform real-time calibration and analysis on the acquired light control analysis image based on the light control standard image, determine whether the light control analysis image meets the light control standard image, and obtain a corresponding calibration result; The setting method of light control accessories includes: The reference parts are set according to the information of each metal accessory; the part information of each reference part is identified, and the part information of each reference part is summarized to set the light control accessory; and the light control accessory is installed in the detection area.

2. The metal furniture quality inspection visualization system based on artificial intelligence according to claim 1 is characterized in that: Methods for adjusting the detection area according to metal fittings information include: According to the information of each metal accessory, a corresponding simulation part is set, and the image of the simulation part is captured under preset conventional lighting conditions and re-inspection lighting conditions to obtain a conventional capture simulation image and a re-inspection capture simulation image; Perform image recognition on the conventional acquisition simulation image and the re-inspection acquisition simulation image respectively, determine the area where the image recognition is abnormal, and mark it as the abnormal simulation area; determine the conventional simulation range and the re-inspection simulation range according to each abnormal simulation area in the conventional acquisition simulation image and the re-inspection acquisition simulation image respectively, and set the initial detection area according to the conventional simulation range and the re-inspection simulation range; The size information of the metal accessory information is identified, and the initial detection area is adjusted according to the size information to obtain the detection area.

3. The metal furniture quality inspection visualization system based on artificial intelligence according to claim 1 is characterized in that: When the image acquisition module receives the light control verification notification sent by the light source control module, it identifies the corresponding accessory acquisition image, marks it as an abnormal accessory acquisition image, removes the abnormal accessory acquisition image, and re-acquires the corresponding accessory acquisition image.

4. The metal furniture quality inspection visualization system based on artificial intelligence according to claim 1 is characterized in that: Methods for performing appearance inspection on metal accessories based on grayscale inspection images include: Identify each color block in the metal accessories, segment the grayscale detection image according to the boundary contour and each color block, and obtain the grayscale area image corresponding to each color block; Identify the grayscale value of each pixel in the grayscale area image; calculate the location set of each pixel, perform pixel evaluation based on each location set to obtain each abnormal pixel; determine the appearance evaluation result based on the distribution of each abnormal pixel.

Images