Image recognition-based automobile seat defect detection method

Abstract

The present application relates to the technical field of car seat surface defect detection, and comprises a kind of automobile seat defect detection method based on image recognition.The method comprises the following steps: S1, image acquisition and modeling;S2, texture continuity detection;S3, degree of swelling judgment and abnormal reason judgment of swelling;S4, stitch correlation analysis;S5, temperature comprehensive judgment;The present application constructs point cloud model to detect the continuity of texture, and executes material filling judgment or stitch correlation analysis when swelling;And temperature comprehensive judgment is carried out in the extension area of stitch, to identify the specific abnormal reason.The present application realizes the automatic identification and classification judgment of seat surface damage, swelling and other defects based on three-dimensional point cloud and texture continuity analysis;Fusion stitch process characteristics and temperature data effectively distinguish material abnormality and structural abnormality;It is suitable for high-precision detection of multi-functional area, effectively reduces manual intervention and misjudgment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of seat detection data processing, and particularly relates to a vehicle seat defect detection method based on image recognition. BACKGROUND

[0002] With the continuous development of the automobile manufacturing industry, the defect detection technology of automobile seats has experienced evolution from traditional manual methods to modern intelligent systems.

[0003] Early, seat defect detection mainly relies on manual visual inspection and manual measurement, and the detection items include appearance quality, size, compression resilience, wear resistance, aging resistance and flame resistance, etc. However, such methods have low efficiency, strong subjectivity, and are difficult to realize standardization, and are difficult to meet the requirements of modern automobile manufacturing for high efficiency and high precision.

[0004] In order to solve these problems, machine vision technology is introduced into seat defect detection. Through industrial cameras and image processing software, defects on the surface of the seat, such as damage, stains and color difference, can be automatically identified, improving the automation and standardization level of detection. With the development of artificial intelligence technology, deep learning and machine learning algorithms are applied to seat defect detection. By training a multi-feature fusion classifier, different materials and colors of seat materials can be identified, and defects such as damage and stains can be detected, improving the accuracy and adaptability of detection.

[0005] In addition, the application of 3D vision technology and point cloud processing algorithm enables seat defect detection to expand from two-dimensional images to three-dimensional space, which can detect structural defects of the seat, such as bulging and collapse, improving the comprehensiveness and precision of detection.

[0006] In summary, the defect detection technology of automobile seats is developing towards automation, intelligence and high precision, gradually meeting the higher requirements of modern automobile manufacturing for product quality and production efficiency.

[0007] Chinese patent publication number CN118864448A discloses a vehicle seat defect detection method, device and medium based on image recognition; it can be seen that the existing vehicle seat defect detection method has the problem of insufficient correlation between defect images and specific defect types and defect causes, and the large amount of calculation when dealing with complex and atypical defects leads to slow recognition speed, and lacks self-adaptive adjustment mechanism and the ability to adapt to diversified defect scenarios. SUMMARY

[0008] Therefore, the present application provides a vehicle seat defect detection method based on image recognition to overcome the weak defect correlation relationship in the prior art and the lack of self-adaptive adjustment mechanism, and to solve the problem of single feedback information and increased risk of misjudgment and omission.

[0009] To achieve the above object, the application provides a car seat defect detection method based on image recognition, comprising,

[0010] The functional areas of the car seat are divided according to the functional area division standard, and the factory delivery requirement standard of each factory delivery detection parameter corresponding to each functional area is obtained;

[0011] The car seat is powered on;

[0012] The visual image of the powered-on car seat and each recognition information in the visual image are continuously collected within a standard collection period, and the each recognition information is matched with the corresponding point to construct a point cloud structure model of the powered-on car seat;

[0013] The texture on the surface of the car seat is continuously detected to obtain a continuous detection result and a corresponding continuous area;

[0014] If the second continuous detection result is obtained, the drum swelling degree judgment is performed, and the corresponding drum swelling abnormal reason judgment is performed when the drum swelling degree abnormal result is obtained;

[0015] The drum swelling degree abnormal result includes the first result of the drum swelling degree judgment and the third result of the drum swelling degree judgment;

[0016] The material filling comprehensive judgment is performed when the first result of the drum swelling degree judgment is obtained, or the seam line correlation analysis is performed when the third result of the drum swelling degree judgment is obtained;

[0017] If the seam line extension area result is obtained by performing the seam line correlation analysis, the temperature data of the powered-on car seat is analyzed based on the point cloud structure model, the temperature comprehensive judgment is performed, the category to which the temperature comprehensive judgment result belongs is determined, and the identification self-checking step is performed when the temperature comprehensive judgment result is the equipment imbalance type drum swelling abnormal category to determine the drum swelling abnormal reason.

[0018] Further, the construction of the point cloud structure model of the powered-on car seat comprises,

[0019] The visual image of the powered-on car seat is continuously collected within a standard collection period;

[0020] The visual image includes a color image and a depth image, and the color image and the depth image are converted into a three-dimensional point cloud;

[0021] Each recognition information in the visual image is obtained and converted into each recognition field, and the recognition field of each point is superimposed and stored in each corresponding point cloud structure;

[0022] The point cloud is divided into each functional area based on the functional area division standard, and the functional area corresponding to each point cloud structure is superimposed and stored in each corresponding point cloud structure in the form of a marker.

[0023] Further, the continuous detection of the texture of the surface of the automobile seat comprises,

[0024] Obtaining a reflection image of the surface of the automobile seat under the irradiation of the movable light source, and dividing the reflection image into a plurality of reflection image blocks based on a reflection image block division standard;

[0025] Traversing each reflection image block, obtaining an actual texture angle in a neighboring reflection image block having a common edge relationship with the reflection image block;

[0026] Obtaining a difference value of the actual texture angle change between the reflection image block and the neighboring reflection image block, and comparing the difference value with a preset texture angle change threshold to obtain a continuous detection result;

[0027] If a first continuous detection result is obtained, there is a damage defect in the reflection image block, and a next damage reminder is triggered;

[0028] If a second continuous detection result is obtained, there is no damage defect in the reflection image block, and a bulging degree judgment is performed.

[0029] Further, the bulging degree judgment comprises,

[0030] Obtaining height information of a sampling point in each functional area based on the point cloud data, and comparing the height information of the sampling point with a standard height interval of the functional area where the sampling point is located to obtain a height comparison result;

[0031] Obtaining a bulging degree judgment result corresponding to each height comparison result, and a category to which the bulging degree judgment result belongs;

[0032] If a first bulging degree judgment result is obtained, a bulging area is stored in each corresponding point cloud structure in the form of a mark, and a material filling comprehensive judgment is performed;

[0033] If a third bulging degree judgment result is obtained, a seam line correlation analysis is performed.

[0034] Further, the material filling comprehensive judgment comprises,

[0035] The material filling comprehensive judgment comprises a material detection step and a filling adjustment step;

[0036] The material detection step is used to obtain whether the material of the functional area corresponding to the sampling point has a performance fatigue condition;

[0037] If the material detection result is a material performance fatigue result, a material replacement reminder is triggered;

[0038] If the material detection result is a material performance stable result, the filling adjustment step is performed;

[0039] Obtain the filler quality in the factory detection parameter corresponding to the functional area, trigger the adjustment filler quality reminder.

[0040] Further, the suture correlation analysis includes,

[0041] Obtain the bulging area surrounded by each sampling point of the bulging mark, and extract the sampling point at the geometric center of the bulging area and the coordinates of each sampling point on the outline of the bulging area;

[0042] Obtain the suture path segment of the functional area where the sampling point at the geometric center of the bulging area is located and the corresponding suture path segment constraint area;

[0043] Draw a perpendicular line with the sampling point at the geometric center as the starting point, which is perpendicular to the suture path segment and has the shortest distance to the line segment constraint area;

[0044] Obtain the intersection of the outline of the bulging area and the perpendicular line as the sampling point, and store the judgment sampling point in the form of a mark to the point cloud structure corresponding to the sampling point;

[0045] Determine whether the judgment sampling point is within the suture path segment constraint area;

[0046] If the judgment sampling point is within the suture path segment constraint area, the bulging area is a suture affected area, and the suture process and support judgment are executed;

[0047] If the judgment sampling point is not within the suture path segment constraint area, the bulging area is a suture extension area, and the temperature comprehensive judgment is executed.

[0048] Further, the suture process and support judgment includes suture process identification, wherein,

[0049] Obtain the suture depth and suture shape of the suture path segment factory detection;

[0050] Obtain the overlapping part of the suture affected area and the suture path segment constraint area, and store the overlapping area in the form of a mark to the point cloud structure corresponding to each sampling point in the overlapping area;

[0051] Obtain the curvature of each sampling point in the overlapping area, and traverse the curvature of each sampling point to obtain the curvature of the sampling point having an adjacent relationship with the sampling point;

[0052] Obtain the actual curvature change difference between the sampling point and the adjacent sampling point, and compare it with the preset curvature change threshold, and obtain the corresponding suture process identification result based on the curvature comparison result;

[0053] If the first curvature comparison result is obtained, the suture support judgment is executed according to the suture process identification result;

[0054] If the second curvature ratio comparison result is obtained, a corresponding stack seam process identification result is triggered to remind the material stack;

[0055] Further, the seam process and support judgment further includes seam support judgment, wherein,

[0056] The end points of the seam path line segment are obtained, and the three-dimensional coordinates of the sampling points at the end points in the point cloud structure and the corresponding actual depth values are extracted;

[0057] The standard seam depth value corresponding to the functional area where the end point is located is extracted, the actual seam height difference value is calculated and compared with the standard seam height difference threshold value;

[0058] Wherein, the actual seam height value is the absolute value of the difference between the standard seam depth value and the actual depth value;

[0059] If the first seam height difference value comparison result is obtained, a lock seam connection failure reminder is triggered;

[0060] If the second seam height difference value comparison result is obtained, an adjustment filler reminder is triggered;

[0061] Further, the temperature comprehensive judgment includes real-time working environment temperature interference exclusion, wherein,

[0062] Determine whether the real-time working environment temperature and the preset standard working environment temperature are both in the linear change interval of the temperature-expansion curve;

[0063] If the two environment temperatures are both in the linear change interval of the temperature-expansion curve, the linear compensation calculation method is used to eliminate the drumming influence caused by the real-time working environment temperature, and after elimination, the drumming temperature correlation judgment is executed;

[0064] If any of the environment temperatures is not in the linear change interval of the temperature-expansion curve, an environment temperature interference reminder is triggered, and after the real-time working environment temperature returns to the standard working environment temperature, the drumming temperature correlation judgment is executed.

[0065] Further, the temperature comprehensive judgment further includes drumming temperature correlation judgment, wherein,

[0066] Obtain each collection point in the direct heating wire affected area corresponding to each functional area, and store the direct heating wire affected area in the form of a mark in the point cloud structure corresponding to the collection point;

[0067] Iterate through each collection point in the direct heating wire affected area to determine whether there is a mark of the drumming area;

[0068] If the bulging area mark exists, the maximum height information in the sampling points of the bulging area mark is obtained and compared with the standard heating height threshold;

[0069] If the maximum height information is greater than or equal to the standard heating height threshold, an abnormal position of the heating wire is triggered;

[0070] If the maximum height information is less than the standard heating height threshold, the powered seat is normal;

[0071] If the bulging area mark does not exist, an adjustment of the filler quality is triggered.

[0072] Compared with the prior art, the beneficial effects of the present application are that the functional areas of the automobile seat are accurately defined by the functional area division standard, the continuous acquisition of the visual image of the powered automobile seat and the construction of the point cloud structure model are realized, the dynamic detection and defect identification of the seat surface texture are supported, the two cases of bulging and sinking are subdivided, the material filling comprehensive judgment and the seam line correlation analysis are respectively adopted, the unbalanced type and the heat source interference type of the bulging abnormality are combined with the temperature data and the temperature comprehensive judgment, and the imbalance type and the heat source interference type of the bulging abnormality are further introduced into the self-checking step to ensure the diagnosis accuracy. The overall method is systematic, comprehensive, can accurately locate the defect cause, improves the accuracy and intelligent level of detection, and effectively guarantees the quality and safety performance of the automobile seat.

[0073] Further, the three-dimensional point cloud is realized by continuously acquiring the color image and the depth image and converting them into three-dimensional point clouds; at the same time, the recognition field and the functional area mark are superimposed to form a point cloud model with clear structure and rich information, which provides an accurate spatial basis and functional area division support for subsequent defect detection and analysis, and improves the detection accuracy and efficiency.

[0074] Further, the movable light source reflection image is divided into a plurality of reflection image blocks, the damage defect is judged based on the texture angle change difference, and the continuous texture of the seat surface is detected with high sensitivity; the damage defect and the normal texture are strictly distinguished by the threshold, which effectively guarantees the accuracy of the detection result and improves the identification ability of the micro surface defect.

[0075] Further, the height information of the sampling point is obtained based on the point cloud data and compared with the standard height interval to realize the quantitative judgment of the bulging degree; different processing strategies are adopted for the bulging degree of sinking and bulging, and the functional area defect is stored combined with the mark to improve the accuracy and refinement of defect positioning, which is convenient for subsequent targeted analysis. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 The flowchart of the automobile seat defect detection method based on image recognition of the embodiment of the present application;

[0077] Figure 2 The logic determination diagram for continuously detecting the texture of the surface of the automobile seat according to the embodiment of the present application is shown in the figure;

[0078] Figure 3 The logic determination diagram for the suture correlation analysis according to the embodiment of the present application is shown in the figure;

[0079] Figure 4 The logic determination diagram for the real-time working environment temperature interference elimination according to the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0080] In order to make the objects and advantages of the present application clearer, the present application will be further described in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0081] The preferred embodiments of the present application will be described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present application and are not used to limit the protection scope of the present application.

[0082] It should be noted that, in the description of the present application, the terms indicating the direction or position relationship such as "upper", "lower", "left", "right", "inner", "outer" and the like are based on the direction or position relationship shown in the drawings, which is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application.

[0083] In addition, it should also be noted that, in the description of the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connection" should be understood in a broad sense, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the communication inside two elements. Those skilled in the art can understand the specific meaning of the above terms in the present application according to the specific circumstances.

[0084] Please refer to Figure 1 The automobile seat defect detection method based on image recognition according to the embodiment of the present application is shown in the figure, and has the characteristics that it comprises,

[0085] The functional areas of the automobile seat are divided according to the functional area division standard, and the factory delivery requirement standard of the factory delivery detection parameters corresponding to each functional area is obtained;

[0086] The automobile seat is powered on;

[0087] acquire visual images of the powered automobile seat and each identification information in the visual images continuously in a standard acquisition cycle, and match the each identification information with a corresponding point to construct a point cloud structure model of the powered automobile seat;

[0088] continuously detect the texture of the surface of the automobile seat to obtain a continuous detection result and a corresponding continuous region;

[0089] if the second continuous detection result is obtained, perform a drum swelling degree judgment, and when a drum swelling abnormal result is obtained, perform a corresponding drum swelling abnormal reason judgment;

[0090] wherein the drum swelling abnormal result includes a drum swelling degree judgment first result and a drum swelling degree judgment third result;

[0091] when the drum swelling degree judgment first result is obtained, perform a material filling comprehensive judgment, or when the drum swelling degree judgment third result is obtained, perform a seam line correlation analysis;

[0092] if the seam line correlation analysis obtains a seam line extension region result, analyze temperature data of the powered automobile seat based on the point cloud structure model, perform a temperature comprehensive judgment and determine a category to which a temperature comprehensive judgment result belongs, and when the temperature comprehensive judgment result is a device imbalance type drum swelling abnormal category, perform an identification self-checking step to determine the drum swelling abnormal reason.

[0093] each functional region of the automobile seat is accurately defined by a functional region division standard, continuous acquisition of visual images of the powered automobile seat and construction of a point cloud structure model are realized, dynamic detection of the texture of the surface of the seat and defect identification are supported; for drum swelling abnormal conditions, two conditions of deflation and swelling are subdivided, material filling comprehensive judgment and seam line correlation analysis are respectively adopted, temperature data and temperature comprehensive judgment are combined to distinguish between device imbalance type and heat source interference type drum swelling abnormalities, and a self-checking step is further introduced to ensure diagnostic accuracy. The overall method is systematic, comprehensive, can accurately locate defect reasons, improves the accuracy and intelligence level of detection, and effectively guarantees the quality and safety performance of the automobile seat.

[0094] Specifically, the point cloud structure model of the powered automobile seat includes,

[0095] acquiring visual images of the powered automobile seat;

[0096] wherein the visual images include color images and height images, and the color images and the height images are converted into three-dimensional point clouds;

[0097] acquiring color and height identification information corresponding to each point in the visual images, and respectively converting the color and height identification information into corresponding identification values, and storing each identification value in the form of a mark in the corresponding point cloud structure;

[0098] The height identification information is an actual height value of each collection point.

[0099] The point cloud structure is divided into each functional area based on a functional area division standard, and each functional area corresponding to the point cloud structure is stored in the form of a mark in the corresponding point cloud structure.

[0100] In the embodiment, under indoor natural lighting conditions, color images and depth images obtained by structure light or binocular calculation are acquired by using an industrial camera installed above a detection station, continuous acquisition is performed at a frame rate of 15 frames per second, the image resolution is set to 1920*1080, and the image acquisition time is uniformly set to a state after the seat is powered on for 5 seconds.

[0101] The color images and the depth images are jointly processed by OpenCV and PCL libraries; each pixel point in the depth image is converted into a three-dimensional space coordinate (X, Y, Z) through perspective projection and is bound with the RGB value of the corresponding pixel in the color image to form a preliminary point cloud; the XYZRGB field structure is adopted, and the number of single-frame point clouds is about 400,000 points.

[0102] The reference planes of each functional area are fitted from the point cloud by using a RANSAC algorithm and a corresponding factory requirement standard;

[0103] The reference plane is a non-filled seat structure surface when the bare chair is wrapped with a skin material and is not filled with materials.

[0104] For each sampling point, the normal distance of the sampling point to the fitted plane is calculated as the actual height value of the point.

[0105] Meanwhile, the color value of the point is extracted and converted into an HSV space value as a material identification parameter.

[0106] The above two values are stored in the point cloud structure.

[0107] In the embodiment, the division of the seat into functional areas includes a cushion area, a lumbar support area, a neck support area, a side wing support area, and a leg support area.

[0108] Three-dimensional data reconstruction of the automobile seat is realized by continuously acquiring color images and depth images and converting them into three-dimensional point clouds; meanwhile, identification fields and functional area marks are superimposed to form a point cloud model with clear structure and rich information, which provides an accurate spatial basis and functional area division support for subsequent defect detection and analysis, and improves the detection accuracy and efficiency.

[0109] Referring to Figure 2 Fig. 2 shows a logic judgment diagram for continuous detection of the texture of the surface of the automobile seat according to the embodiment of the application;

[0110] Specifically, the continuous detection of the texture of the surface of the automobile seat comprises,

[0111] an image of reflection of the surface of the automobile seat under irradiation of the movable light source is acquired, and the image of reflection is divided into a plurality of image blocks of reflection based on a division standard of image blocks of reflection;

[0112] each image block of reflection is traversed, and an actual texture angle in a neighboring image block of reflection having a common edge relationship with the image block of reflection is acquired;

[0113] a difference value of change of the actual texture angle between the image block of reflection and the neighboring image block of reflection is obtained and compared with a standard threshold value of change of texture angle, so as to obtain a continuous detection result;

[0114] if the first continuous detection result is obtained, there is a damage defect in the image block of reflection, and a damage reminder is triggered;

[0115] if the second continuous detection result is obtained, there is no damage defect in the image block of reflection, and a degree of bulging is judged.

[0116] In the embodiment, after the point cloud model is constructed, the surface of the electrified automobile seat is imaged under irradiation of the configured movable light source, and a high-resolution industrial camera or a structured light scanner is used to acquire an image of reflection of the surface. The irradiation time of the movable light source is 5 seconds.

[0117] The image of reflection is a gray image.

[0118] Based on a division standard of image blocks of reflection, the image of reflection is divided into a plurality of image blocks of reflection with each block of 32x32 pixels.

[0119] An image processing algorithm is used for each image block of reflection to acquire an actual texture angle θ of the region.

[0120] The actual texture angle is in degrees and ranges from 0 to 180.

[0121] Each image block of reflection is traversed, and a neighboring image block having a common edge relationship is compared, a difference value of change of the actual texture angle is calculated, and a comparison with a standard threshold value of change of texture angle is performed.

[0122] If the difference value of change of the actual texture angle is greater than the standard threshold value of change of texture angle, the texture is suddenly changed, there may be a broken line or damage, and the first continuous detection result is obtained.

[0123] If the difference value of change of the actual texture angle is less than or equal to the standard threshold value of change of texture angle, the texture is continuously changed, and the second continuous detection result is obtained.

[0124] The standard threshold value of change of texture angle is 25 degrees according to historical factory identification data.

[0125] The movable light source reflection image is divided into several reflection image blocks, the damage defect is judged based on the texture angle change difference value, and the high-sensitivity detection of the continuous texture on the seat surface is realized; the damage defect and the normal texture are strictly distinguished by a threshold value, the accuracy of the detection result is effectively guaranteed, and the recognition ability of the micro surface defect is improved.

[0126] Specifically, the execution of the bulging degree judgment includes,

[0127] Based on the point cloud data, the actual height value corresponding to the actual height information of any sampling point in each functional area is obtained, and the actual height value is compared with the standard actual height interval of the functional area where the sampling point is located to obtain a height comparison result;

[0128] Obtain the bulging degree judgment result corresponding to each height comparison result, and the category to which the bulging degree judgment result belongs;

[0129] If the first bulging degree judgment result is obtained, the bulging area is stored in the form of a marker in each corresponding point cloud structure, and a material filling comprehensive judgment is performed;

[0130] If the third bulging degree judgment result is obtained, perform a suture correlation analysis.

[0131] In this embodiment, under the condition of selecting artificial leather fabric, the standard actual height interval of each area is:

[0132] The standard actual height interval of the cushion area is [0, 2], the standard actual height interval of the waist rest area is [1.5, 3], the standard actual height interval of the neck support area is [0.5, 2.5], the standard actual height interval of the side wing support area is [1, 1.5], and the standard actual height interval of the leg support area is [0, 3], each interval unit is mm;

[0133] Wherein, each interval unit is millimeter, the interval floating value is the floating value of the sampling point on the normal vector of the corresponding reference plane of the corresponding functional area, and the direction outward to the seat is positive;

[0134] The actual height value is compared with the standard actual height interval of the functional area where the sampling point is located;

[0135] If the actual height value is less than the minimum value of the standard actual height interval, the first bulging degree judgment result is obtained, the actual bulging degree is less than the normal interval, and a material filling comprehensive judgment is performed;

[0136] If the actual height value is within the standard actual height interval, the second bulging degree judgment result is obtained, and the actual bulging degree is within the normal interval;

[0137] If the actual height value is greater than the maximum value of the standard actual height interval, a third result of the bulging degree judgment is obtained, the actual bulging degree is greater than the normal interval, and a suture correlation analysis is performed.

[0138] Based on the point cloud data, the height information of the sampling points is obtained and compared with the standard height interval to realize quantitative judgment of the bulging degree. Different processing strategies are adopted for the bulging degree of collapse and bulging, and the functional area defects are stored by marking, which improves the accuracy and refinement of defect positioning and facilitates subsequent targeted analysis.

[0139] Specifically, the material filling comprehensive judgment includes,

[0140] The material filling comprehensive judgment includes a material detection step and a filling adjustment step.

[0141] The material detection step is used to obtain whether the material of the functional area corresponding to the sampling point has performance fatigue;

[0142] If the material detection result is a material performance fatigue result, an adjustment material reminder is triggered;

[0143] If the material detection result is a material performance stability result, a filling adjustment step is performed;

[0144] Through the filling adjustment step, the filler quality in the factory detection parameter corresponding to the functional area is obtained, and an adjustment filler quality reminder is triggered.

[0145] In this embodiment, under the irradiation of the configured movable light source, the surface of the powered automobile seat is imaged to obtain a surface reflection image using a high-resolution industrial camera or a structured light scanner, and the gloss reflection intensity of each collection point in the surface reflection image is obtained. The irradiation time of the movable light source is 5 seconds;

[0146] The gloss reflection intensity of all collection points in each functional area is obtained, the average value of the gloss reflection of each functional area is calculated, and the average value is compared with the standard reflection intensity;

[0147] The average value of the gloss reflection is calculated as follows:

[0148]

[0149] Wherein, is the average value of the gloss reflection;

[0150] N is the number of sampling points of the functional area;

[0151] is the reflection intensity of a single sampling point;

[0152] Define the gloss attenuation rate:

[0153]

[0154] wherein, is the standard reflection intensity;

[0155] The standard reflection intensity under the factory requirement standard is obtained, and in this embodiment, the standard reflection intensity under the artificial leather fabric is 0.25;

[0156] The average value of the gloss reflection is compared with the standard reflection intensity;

[0157] If the average value of the gloss reflection is greater than the standard reflection intensity, the reflection is significantly reduced, and a material fatigue reminder is triggered;

[0158] If the average value of the gloss reflection is less than or equal to the standard reflection intensity, it is judged that the gloss is normal, that is, the filler quality under the factory requirement standard is small, and a filler quality adjustment reminder is triggered.

[0159] Through the two-step comprehensive judgment of material detection and filling adjustment, material fatigue is found in time and a replacement reminder is triggered, so as to ensure the stable performance of the seat material. The filling adjustment step guides the adjustment of the filler quality according to the factory detection parameters, improves the scientificity and rationality of the filling effect, prolongs the service life of the seat, and improves the riding comfort.

[0160] Referring to Figure 3 , which is a logical determination diagram of the suture correlation analysis described in the embodiment of the present application;

[0161] Specifically, the suture correlation analysis includes,

[0162] Obtain the bulging area formed by each sampling point with a bulging mark, and extract the coordinates of the collection point at the geometric center of the bulging area and each sampling point on the contour of the bulging area;

[0163] Obtain the suture path line segment of the functional area where the collection point at the geometric center of the bulging area is located and the corresponding suture path line segment constraint area;

[0164] Take the collection point at the geometric center as the starting point, draw a plurality of candidate perpendicular lines perpendicular to the suture path line segment, and obtain the target perpendicular line with the shortest distance to the line segment constraint area;

[0165] Obtain the intersection point of the contour of the bulging area and the target perpendicular line as a judgment collection point, and store the judgment collection point in the form of a mark to the point cloud structure corresponding to the collection point;

[0166] Determine whether the judgment collection point is within the suture path line segment constraint area;

[0167] If the judgment collection point is within the suture path line segment constraint area, the bulging area is a suture affected area, and the suture process and support judgment are performed;

[0168] If it is judged that the collection point is not in the suture path line segment constraint area, the bulging area is the suture extension area, and the temperature comprehensive judgment is performed.

[0169] In this embodiment, the suture path line segment constraint area is the area where the suture is deformed due to the actual wiring track and the deformation of the two sides due to tension, restraint and sewing traction;

[0170] The suture extension area is a non-suture path line segment constraint area.

[0171] The relationship between the bulging area and the suture path is determined by combining the geometric center and the contour sampling point, the suture affected area and the extension area are distinguished, the suture process judgment or the temperature comprehensive judgment is performed for different areas, the accurate positioning of the bulging abnormality is realized, and the pertinence and scientificity of defect analysis are improved.

[0172] Specifically, the suture process and support judgment includes suture process identification, wherein,

[0173] Obtain the suture height and suture shape of the suture path line segment out-of-factory detection;

[0174] Obtain the overlapping part of the suture affected area and the suture path line segment constraint area, and store the overlapping area in the form of marks to the point cloud structure corresponding to each collection point of the overlapping part;

[0175] Obtain the curvature of each collection point in the overlapping area, traverse the curvature of each collection point, and obtain the curvature of the collection point having an adjacent relationship with the collection point;

[0176] Obtain the actual curvature change difference between the collection point and the adjacent collection point, and compare it with the preset curvature change threshold, and obtain the corresponding suture process identification result based on the curvature comparison result;

[0177] If the first curvature comparison result is obtained, the corresponding suture process identification result is stretched, and the suture support judgment is performed;

[0178] If the second curvature comparison result is obtained, the corresponding suture process identification result is stacked, and the material stacking reminder is triggered.

[0179] In this embodiment, the curvature of each sampling point in the overlapping area is obtained , and the curvature of the adjacent point and is obtained and ;

[0180] The formula for calculating the actual curvature change difference is:

[0181]

[0182] And compare it with the preset curvature change threshold;

[0183] The preset curvature change thresholds for each functional area are as follows: 0.15 for the seat cushion area, 0.2 for the lumbar support area, 0.2 for the neck support area, 0.1 for the side support area, and 0.15 for the leg rest area.

[0184] If the actual curvature change difference is less than or equal to the preset curvature change threshold, the first curvature comparison result is obtained, which corresponds to the stretching suture process identification result, and suture support judgment is performed.

[0185] If the actual curvature change difference is greater than the preset curvature change threshold, a second curvature comparison result is obtained, which corresponds to the stacked stitching process identification result, triggering a material stacking reminder.

[0186] Based on factory inspection standards for stitch depth and shape, and combined with curvature change comparison, the stitching process type is identified, distinguishing between stretched stitches and stacked stitches. Stacked stitches trigger material stacking alerts, scientifically identifying stitching process defects, supporting efficient and precise process quality control, and ensuring the stability of seat manufacturing processes.

[0187] Specifically, the suture process and support judgment also include suture support judgment, wherein,

[0188] Obtain the endpoints of the suture path segment, and extract the three-dimensional coordinates of the sampling points at the endpoints in the point cloud structure and their corresponding actual height values;

[0189] Extract the standard suture height value corresponding to the factory requirements standard of the functional area where the endpoint is located, calculate the actual suture height difference, and compare the actual suture height difference value with the standard suture height difference threshold.

[0190] The actual suture height value is the absolute value of the difference between the standard suture height value and the actual height value.

[0191] If the first seam height difference comparison result is obtained, a lock seam connection fault alert will be triggered.

[0192] If the comparison result of the second suture height difference is obtained, a reminder to adjust the filler quality will be triggered.

[0193] In this embodiment, the endpoints of the suture path segments are identified visually;

[0194] The endpoints include the start and end points of the suture path segment;

[0195] Calculate the actual suture height difference ;

[0196] wherein, is a standard stitch height value;

[0197] is an actual height value;

[0198] The standard stitch height value under the factory requirement standard in the embodiment is 1.8 mm; the standard stitch height difference threshold is 0.3 mm;

[0199] The actual stitch height difference value is obtained and compared with the standard stitch height difference threshold;

[0200] If the actual stitch height difference value is greater than the standard stitch height difference threshold, a first stitch height difference value comparison result is obtained, and there is a lock stitch connection failure;

[0201] If the actual stitch height difference value is less than or equal to the standard stitch height difference threshold, a second stitch height difference value comparison result is obtained, and the quality of the filler is small;

[0202] The lock stitch connection failure is that the face thread and the bottom thread of the stitch do not form a buckling structure, that is, they are not crossed and wound in the middle of the material, resulting in a failure that the structure at the suture is loose or invalid.

[0203] By comparing the three-dimensional coordinates of the stitch path end points and the actual depth value with the standard stitch depth value, the stitch height difference is quantified, and intelligent prompting of the lock stitch connection defect and the filler adjustment is realized; the method refines the stitch support quality judgment, which helps to discover key structural defects in the manufacturing process in time and improves the overall durability of the seat.

[0204] Referring to Figure 4 , which is a logic determination diagram for real-time working environment temperature interference elimination of the embodiment of the present application;

[0205] Specifically, the temperature comprehensive judgment includes real-time working environment temperature interference elimination, wherein,

[0206] It is judged whether the real-time working environment temperature and the preset standard working environment temperature are both in the linear change interval of the temperature-expansion curve;

[0207] If the two environment temperatures are both in the linear change interval of the temperature-expansion curve, a linear compensation calculation method is used to eliminate the drumming influence caused by the real-time working environment temperature, and after elimination, the drumming temperature correlation judgment is performed;

[0208] If any of the environment temperatures is not in the linear change interval of the temperature-expansion curve, an environment temperature interference prompt is triggered, and after the real-time working environment temperature returns to the standard working environment temperature, the drumming temperature correlation judgment is performed.

[0209] In this embodiment, the temperature-expansion curve refers to a graph showing the change trend of the thermal expansion of the structure of a material under different temperature conditions. The curve usually has a two-section structure, i.e., a linear change interval and a nonlinear change interval.

[0210] The linear compensation calculation method is to determine whether the real-time working environment temperature is greater than the preset standard working environment temperature, and subtract the bulging value caused by the difference between the real-time working environment temperature and the preset standard working environment temperature when the real-time working environment temperature is greater than the preset standard working environment temperature.

[0211] The temperature difference and the expansion trend are analyzed to ensure the accuracy of the bulging degree determination, which avoids the misleading of the detection result caused by the fluctuation of the environmental temperature, and improves the robustness and practicality of the detection method.

[0212] Specifically, the temperature comprehensive judgment further includes a bulging temperature correlation judgment, wherein,

[0213] Each collection point in the direct influence area of the heating wire corresponding to each functional area is obtained, and the direct influence area of the heating wire is stored in the point cloud structure corresponding to each collection point in the form of a mark;

[0214] Each collection point in the direct influence area of the heating wire is traversed to determine whether there is a mark of the bulging area;

[0215] If there is a mark of the bulging area, the maximum height information of the collection points with the mark of the bulging area is obtained and compared with a standard heated height threshold value;

[0216] If the maximum height information is greater than or equal to the standard heated height threshold value, an adjustment of the position of the heating wire is triggered;

[0217] If the maximum height information is less than the standard heated height threshold value, the powered seat is normal;

[0218] If there is no mark of the bulging area, an adjustment of the quality of the filler is triggered.

[0219] In this embodiment, the center line of the heating wire in the area is obtained, and a direct influence area of the heating wire is constructed with a radius of 1.5 cm;

[0220] The maximum height information of the collection points with the mark of the bulging area is obtained and compared with a standard heated height threshold value;

[0221] The standard heated height threshold values corresponding to each functional area are as follows: a seat cushion area standard heated height threshold value of 4 mm, a lumbar support area standard heated height threshold value of 6 mm, a neck support area standard heated height threshold value of 5 mm, a side wing support area standard heated height threshold value of 3 mm, and a leg support area standard heated height threshold value of 6 mm.

[0222] If the maximum height information is greater than the standard heating height threshold, the local heat of the heating wire is too high, triggering an adjustment of the heating wire position reminder;

[0223] If the maximum height information is less than or equal to the standard heating height threshold, the heating wire has no effect, and the system will trigger an adjustment of the filler quality reminder to check the filler quality.

[0224] In combination with the point cloud marking of the area directly affected by the heating wire and the bulging area marking, quantitative judgment of the bulging temperature correlation is realized. For abnormal height, a position abnormality reminder is triggered, and for no abnormal condition, the seat is determined to be normal, improving the accuracy and real-time response capability of temperature-related defect detection and ensuring the safe and stable operation of the seat.

[0225] Thus, the technical solutions of the present application have been described in combination with the preferred embodiments shown in the accompanying drawings, but those skilled in the art can easily understand that the protection scope of the present application is obviously not limited to these specific embodiments. Those skilled in the art can make equivalent changes or replacements to the related technical features without departing from the principles of the present application, and the technical solutions after such changes or replacements will fall within the protection scope of the present application.

[0226] The above description is only the preferred embodiments of the present application and is not intended to limit the present application; for those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting defects in automobile seats based on image recognition, characterized in that, include, The functional areas of the car seat are defined according to the functional area division standard, and the factory requirement standards of each factory test parameter corresponding to each functional area are obtained. Power on the car seat; In the current detection cycle, visual images of the electric car seat and various identification information within the visual images are continuously acquired, and the identification information is matched with corresponding points to construct a point cloud structure model of the electric car seat. The texture of the car seat surface is continuously detected to obtain a reflection image of the car seat surface under the illumination of a movable light source. The reflection image is then divided into several reflection image blocks based on the reflection image block division standard. If the first consecutive detection result is obtained, there is a defect in the reflective image block, triggering a damage alert; If the second consecutive detection result is obtained, and there are no defects or flaws within the reflected image block, then a bulge degree judgment is performed; The process of determining the degree of bulging includes obtaining the height information of the sampling points based on point cloud data and comparing it with the standard height range to achieve a quantitative determination of the degree of bulging, and performing the corresponding determination of the cause of the bulging anomaly when an abnormal result of the degree of bulging is obtained. The abnormal swelling result includes a first swelling judgment result and a third swelling judgment result; When the first result of the swelling degree judgment is obtained, a comprehensive judgment of the material filling is performed, wherein the actual swelling degree is less than the normal range; or when the third result of the swelling degree judgment is obtained, a suture correlation analysis is performed, wherein the actual swelling degree is greater than the normal range. The material filling comprehensive judgment includes a material detection step and a filling adjustment step; the material detection step is used to determine whether the material in the functional area corresponding to the sampling point is experiencing performance fatigue. The suture correlation analysis includes, Obtain the bulging region formed by the sampling points with bulging marks, and extract the sampling point at the geometric center of the bulging region and the coordinates of each sampling point on the outline of the bulging region; Obtain the suture path segment and the corresponding suture path segment constraint area of ​​the functional area where the collection point is located at the geometric center of the bulging area. Starting from the sampling point at the geometric center, draw several candidate perpendicular lines perpendicular to the suture path segment, and obtain the target perpendicular line that is shortest to the constrained area of ​​the line segment. The intersection of the bulging region contour and the target vertical line is obtained as the judgment collection point, and the judgment collection point is superimposed and stored in the point cloud structure corresponding to the judgment collection point in the form of a mark. Determine whether the sampling point is within the constrained area of ​​the suture path segment; If it is determined that the sampling point is within the constrained area of ​​the suture path segment, and the bulging area is the area affected by the suture, then the suture process and support judgment are performed. If it is determined that the sampling point is not within the constrained area of ​​the suture path segment, and the bulging area is the outer extension area of ​​the suture, then a comprehensive temperature judgment is performed. If the suture correlation analysis yields the results of the suture extension area, the temperature data of the electric car seat is analyzed based on the point cloud structure model. A comprehensive temperature judgment is performed to determine the category to which the comprehensive temperature judgment result belongs. If the comprehensive temperature judgment result is classified as a device imbalance-type bulging abnormality, a self-checking step is performed to determine the cause of the bulging abnormality.

2. The method for detecting defects in automobile seats based on image recognition according to claim 1, characterized in that, The construction of the point cloud structure model of the electric car seat includes... Acquire visual images of powered car seats; The visual image includes a color image and a height image, which are then converted into a three-dimensional point cloud. The color and height recognition information corresponding to each point in the visual image are obtained and converted into corresponding recognition values. The recognition values ​​are then superimposed and stored in the corresponding point cloud structure in the form of tags. The height identification information refers to the actual height value of each collection point; Based on the functional region division standard, the point cloud structure is divided into various functional regions, and the functional regions corresponding to each point cloud structure are superimposed and stored in the corresponding point cloud structure in the form of tags.

3. The method for detecting defects in automobile seats based on image recognition according to claim 1, characterized in that, The continuous detection of the texture on the surface of the car seat includes, Acquire a reflection image of the car seat surface under illumination by a movable light source, and divide the reflection image into several reflection image blocks based on a reflection image block division standard; Traverse each reflection image block and obtain the actual texture angle within adjacent reflection image blocks that share an edge with the reflection image block; The actual texture angle change difference between the reflected image patch and its adjacent reflected image patches is obtained and compared with the standard texture angle change threshold to obtain continuous detection results.

4. The method for detecting defects in automobile seats based on image recognition according to claim 3, characterized in that, The determination of the degree of bulging includes, Based on point cloud data, the actual height value corresponding to the actual height information of any sampling point in each functional area is obtained, and the actual height value is compared with the standard actual height range of the functional area where the sampling point is located to obtain the height comparison result. Obtain the bulging degree judgment result corresponding to each height comparison result, and the category to which the bulging degree judgment result belongs; If the first result of the swelling degree judgment is obtained, the swelling area is superimposed and stored in the corresponding point cloud structure in the form of a mark, and the material filling comprehensive judgment is performed. If the third result of the swelling degree judgment is obtained, perform suture correlation analysis.

5. The image recognition-based automotive seat defect detection method according to claim 4, characterized in that, The comprehensive judgment of material filling includes, If the material test result indicates material performance fatigue, a material adjustment reminder will be triggered. If the material test result indicates that the material performance is stable, proceed with the filling adjustment step. The filling adjustment step obtains the filling quality within the factory test parameters corresponding to the functional area, triggering a reminder to adjust the filling quality.

6. The method for detecting defects in automobile seats based on image recognition according to claim 1, characterized in that, The suture process and support determination includes suture process identification, wherein... Obtain the suture height and suture shape from the factory inspection of the suture path segment; Obtain the overlapping portion of the suture influence area and the suture path line segment constraint area, and store the overlapping area in the form of a mark in the point cloud structure corresponding to each acquisition point of the overlapping portion; Obtain the curvature of each sampling point within the overlapping region, traverse the curvature of each sampling point, and obtain the curvature of sampling points that are adjacent to the sampling point. The actual curvature change difference between the sampling point and the adjacent sampling points is obtained and compared with the preset curvature change threshold. Based on the curvature comparison result, the corresponding suture process identification result is obtained. If the first curvature comparison result is obtained, and the corresponding smoothing stitch process identification result is obtained, then the stitch support judgment is performed. If the second curvature comparison result is obtained, the corresponding stacked stitching process identification result will trigger a material stacking reminder.

7. The method for detecting defects in automobile seats based on image recognition according to claim 1, characterized in that, The suture process and support judgment also include suture support judgment, wherein... Obtain the endpoints of the suture path segment, and extract the three-dimensional coordinates of the sampling points at the endpoints in the point cloud structure and their corresponding actual height values; Extract the standard suture height value corresponding to the factory requirements standard of the functional area where the endpoint is located, calculate the actual suture height difference, and compare the actual suture height difference value with the standard suture height difference threshold. The actual suture height value is the absolute value of the difference between the standard suture height value and the actual height value. If the first seam height difference comparison result is obtained, a lock seam connection fault alert will be triggered. If the comparison result of the second suture height difference is obtained, a reminder to adjust the filler will be triggered.

8. The method for detecting defects in automobile seats based on image recognition according to claim 1, characterized in that, The comprehensive temperature judgment includes eliminating interference from the real-time working environment temperature, wherein... Determine whether the real-time operating environment temperature and the preset standard operating environment temperature are both within the linear variation range of the temperature-expansion curve. If both ambient temperatures are within the linear variation range of the temperature-expansion curve, then the linear compensation calculation method is used to remove the bulging effect caused by the real-time working ambient temperature, and the bulging temperature correlation judgment is performed after the removal. If any ambient temperature is not within the linear change range of the temperature-expansion curve, an ambient temperature interference alert will be triggered, and an expansion temperature correlation judgment will be performed after the real-time working ambient temperature recovers to the standard working ambient temperature.

9. The method for detecting defects in automobile seats based on image recognition according to claim 8, characterized in that, The comprehensive temperature judgment also includes a correlation judgment of bulging temperature, wherein... Acquire each collection point in the area directly affected by the heating wire in each functional area, and store the area directly affected by the heating wire in the form of a mark in the point cloud structure corresponding to each collection point; Traverse all sampling points within the area directly affected by the heating wire to determine if there are any markers for bulging areas; If a bulging area is marked, obtain the maximum height information of each collection point with the bulging area mark and compare it with the standard heated height threshold; If the maximum height information is greater than or equal to the standard heating height threshold, a reminder to adjust the heating wire position will be triggered. If the maximum height information is less than the standard heated height threshold, the powered seat is normal. If no marker for the bulging area is found, a reminder to adjust the filler quality will be triggered.

Images