Multi-parameter visual online detection method and device suitable for barbed suture

Abstract

This invention belongs to the field of online inspection technology, specifically a multi-parameter visual online inspection method and device suitable for barbed sutures. The method includes: defining suture inspection parameters; obtaining parameter adjustment data based on material reflectivity and morphological characteristics; adjusting the suture inspection parameters according to the parameter adjustment data to obtain target inspection parameters; acquiring multi-view image data of the current barbed suture; processing the data using a multi-view inspection method to obtain multi-dimensional inspection data; determining whether the multi-dimensional inspection data meets a preset threshold based on the target inspection parameters; outputting a pass / fail result if yes, otherwise marking the defect location type; analyzing the correlation between the defect location type and the suture inspection parameters to obtain defect detection correlation correction data; and adjusting the target inspection parameters to obtain accurate inspection parameters. This invention can adapt to the inspection needs of barbed sutures of various materials and specifications, reducing equipment investment costs and improving inspection accuracy and stability.

Description

Technical Field

[0001] This invention belongs to the field of online inspection technology, specifically a multi-parameter visual online inspection method and device applicable to barbed sutures. Background Technology

[0002] As a critical consumable in surgical procedures, barbed sutures directly impact surgical safety and healing outcomes due to parameters such as suture diameter accuracy, barb morphology, and surface integrity, placing stringent medical-grade requirements on detection precision. However, existing multi-parameter visual online detection methods face bottlenecks in practical applications, severely limiting their reliability and efficiency.

[0003] On the one hand, the diversity of materials used in barbed sutures leads to unstable detection accuracy. Materials widely used in clinical practice, such as nylon, polylactic acid, and silk, vary significantly in reflectivity, transparency, and texture: highly reflective materials easily create light spots that mask defects, transparent and absorbable materials easily allow background penetration, resulting in blurred outlines, while highly textured materials are easily confused with surface defects. Traditional detection methods often use fixed parameter configurations, which cannot dynamically adapt to the optical and physical properties of different materials, leading to high rates of missed and false detections, and measurement errors in critical dimensions exceeding allowable limits, making it difficult to meet the high-precision detection requirements of medical consumables.

[0004] On the other hand, the inefficiency of multi-view data fusion is a prominent issue. The structure of barbed sutures is complex, requiring inspection of multiple areas including the needle tip, needle body, and inner / outer barb rings. Existing technologies often employ a multi-camera independent acquisition mode, but lack a unified synchronization triggering mechanism and data integration logic, leading to asynchronous image acquisition from different perspectives and inconsistent parameter measurement benchmarks. Furthermore, a single perspective easily creates detection blind spots, while fragmented processing of multi-view data can cause information redundancy or conflicts, not only reducing detection efficiency but also affecting the accuracy of defect localization and parameter calculation due to data fusion deviations.

[0005] Furthermore, existing detection methods are mostly static parameter settings, which cannot dynamically optimize detection parameters based on material changes and defect feedback, further exacerbating the instability of detection accuracy. Therefore, there is an urgent need for multi-parameter visual online detection methods and equipment suitable for barbed sutures to address the pain points of existing technologies and ensure the accuracy, stability, and efficiency of barbed suture detection. Summary of the Invention

[0006] To overcome the shortcomings of existing technologies, this invention proposes a multi-parameter visual online detection method and device suitable for barbed sutures. This invention primarily addresses the problems of unstable detection accuracy and inefficient multi-view data fusion in existing barbed suture detection methods due to material differences.

[0007] The technical solution adopted by this invention to solve its technical problem is: the multi-parameter visual online detection method for barbed sutures provided by this invention includes:

[0008] Collect material property data and specification parameters of barbed sutures, formulate suture detection parameters based on target detection accuracy requirements, and use imaging calibration methods to obtain parameter adjustment data based on material reflectivity and morphological characteristics analysis.

[0009] The target detection parameters are obtained by adjusting the suture detection parameters based on the parameter adjustment data. Multi-view image data of the current barbed suture is collected and processed using a multi-view detection method to obtain multi-dimensional detection data.

[0010] Based on the target detection parameters, determine whether the multi-dimensional detection data meets the preset threshold. If yes, output the qualified judgment result; otherwise, mark the defect location type.

[0011] By analyzing the correlation between defect location type and suture detection parameters, defect detection correlation correction data is obtained, and the target detection parameters are adjusted to obtain accurate detection parameters.

[0012] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for determining suture detection parameters:

[0013] The optical, physical, and processing compatibility characteristics of the suture are collected as material property data, and the suture body parameters and barb parameters are collected as specification parameters.

[0014] Accuracy thresholds are established based on the geometric parameters, defects, and morphology of the object being inspected, and inspection efficiency requirements are determined in conjunction with production speed.

[0015] Based on the detection efficiency requirements, the resolution, shooting frequency, field of view and focal length, light source type and light intensity are determined to obtain the image acquisition parameters.

[0016] Based on the accuracy threshold and target defect characteristics, noise reduction parameters, edge enhancement parameters, and illumination correction parameters are selected as processing parameters. Workstations are divided according to the preset detection logic, and the camera trigger interval is set based on the pulse of the main encoder of the production line to obtain the detection process parameters.

[0017] Based on the historical testing parameters and results of similar products, correction logic and process characteristic correction parameters are extracted. The image acquisition parameters, processing parameters, and testing process parameters are adjusted to obtain the suture detection parameters.

[0018] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for analyzing and obtaining parameter adjustment data:

[0019] The average reflectivity, reflectance distribution uniformity, and grayscale difference between the suture and the background are extracted as material reflectance characteristics for the barbed suture.

[0020] With the goal of clearly extracting image characteristics, it determines whether the reflective properties of the current material meet the detection requirements and identifies the type of deviation to obtain the reflective property deviation.

[0021] The sharpness of the barb edge, the consistency of the line shape, the uniformity of the barb distribution, and the density of the surface texture of the barb suture are extracted as morphological features to determine whether they affect the deviation of the morphological features obtained from parameter measurement and defect identification.

[0022] Based on the deviations in reflective properties and morphological characteristics, the adjustment direction and parameters of the corresponding matching suture detection parameters are determined.

[0023] Based on the preset accuracy requirements and adjustment direction, the adjustment coefficients of the adjustment parameters are calculated to obtain the parameter adjustment data.

[0024] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for adjusting the target detection parameters:

[0025] A mapping table for adjusting suture detection parameters and parameter adjustment data is established by mapping the parameters one-to-one.

[0026] Based on the directness of the impact of different adjustment parameters on the detection effect and the response speed, the parameter adjustment priority is obtained by prioritizing them.

[0027] The adjustment operation is performed step by step according to the parameter adjustment priority and the adjustment parameter mapping table. With the target detection accuracy requirement as the constraint, the adjustment range of each adjustment parameter is checked to see if it meets the standard. If it does, the target detection parameter is output; otherwise, the adjustment range is corrected.

[0028] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for obtaining multi-dimensional detection data:

[0029] The multi-view camera array is calibrated according to the target detection parameters, the camera acquisition interval is set, the detection area is divided according to the suture structure, the image is acquired, and the multi-view image data is obtained by storing the data in the format of product number-detection area-camera number-time stamp.

[0030] The image data from multiple perspectives is converted to grayscale, denoised, and illuminated, and then geometrically corrected and cropped to obtain the preliminary processed image data.

[0031] Feature extraction is performed on the pre-processed image data to obtain contour defect features, and geometric dimensions, morphological accuracy, and feature surface defect parameters are calculated to obtain multi-dimensional detection data.

[0032] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for determining whether multi-dimensional detection data meets a preset threshold:

[0033] Based on the target detection parameters, the judgment standard thresholds are set in layers according to parameter type, detection area and defect level.

[0034] The multi-dimensional detection data are compared with the target detection data one by one according to the judgment standard threshold, and the qualified judgment result is output.

[0035] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for marking the defect location type:

[0036] Defect coordinates are determined based on multi-dimensional detection data. Combined with the divided detection areas, the location of the defect is marked, and the corresponding camera view image is associated to mark the defect image area.

[0037] Defect image regions are classified according to geometric parameters, surface defects, and morphology to determine the defect image type.

[0038] The defect image level is labeled according to the defect size and impact of the defect image area, and the defect location type is obtained by classifying it according to product number-non-conformance-time stamp.

[0039] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for obtaining defect detection correlation correction data:

[0040] Standardized defect location types are processed to obtain standard defect data. The suture detection parameters of the corresponding defective products are extracted and matched with the standard defect data according to the product number and detection timestamp.

[0041] The association analysis dimensions are obtained by associating defect type with parameter type, defect location with parameter effective range, and defect index with parameter value deviation.

[0042] Based on the correlation analysis dimensions, parameter value distribution characteristics, defect occurrence rate correlation characteristics, and indicator parameter correlation characteristics are extracted from standard defect data as correlation features.

[0043] Sensitivity is calculated using the defect improvement rate as a sensitivity index. The influence of each suture detection parameter on the associated features is quantified, and parameters that reach the preset sensitivity threshold are selected as high-sensitivity parameters.

[0044] Based on highly sensitive parameters and associated features, parameter correction rules are formulated to obtain defect detection associated correction data.

[0045] The multi-parameter visual online detection method for barbed sutures provided by this invention includes the following steps for adjusting the detection parameters to obtain accurate results:

[0046] Based on the target detection parameters, a one-to-one correspondence is established with the defect detection-related correction data to form a two-way mapping relationship. Scenario correction data is obtained by filtering according to the product material, specifications, and defect type currently being detected.

[0047] Correction priorities are determined based on parameter adjustment priorities and the degree of impact, and the adjustment parameter values ​​corresponding to the correction priorities are calculated based on scenario correction data.

[0048] Based on the parameter detection results and equipment boundaries, a conflict coordination strategy is formulated, and the target detection parameters are adjusted according to the adjusted parameter values ​​to obtain accurate detection parameters.

[0049] The present invention provides a multi-parameter visual online inspection device suitable for barbed sutures, comprising:

[0050] The intelligent parameter optimization module is used to collect material property data and specification parameters of the barbed suture, formulate suture detection parameters in combination with the target detection accuracy requirements, and obtain parameter adjustment data by analyzing the material's reflective properties and morphological characteristics using an imaging calibration method.

[0051] The online image detection module is used to adjust the suture detection parameters according to the parameter adjustment data to obtain the target detection parameters, collect multi-view image data of the current barbed suture, and process it using a multi-view detection method to obtain multi-dimensional detection data.

[0052] The defect determination and marking module is used to determine whether the multi-dimensional detection data meets the preset threshold based on the target detection parameters. If it does, the module outputs a qualified judgment result; otherwise, it marks the defect location type.

[0053] The parameter closed-loop correction module is used to analyze the correlation between defect location type and suture detection parameters to obtain defect detection correlation correction data, and adjust the target detection parameters to obtain accurate detection parameters.

[0054] The beneficial effects of this invention are as follows:

[0055] 1. This invention achieves high accuracy in geometric parameter measurement and improves defect recognition rate through material-morphology adaptive calibration and multi-view collaborative detection. It significantly reduces the false negative and false positive rates for low-contrast and minute defects, meeting medical-grade detection accuracy requirements. Detection parameters can be dynamically adjusted based on the reflective properties, morphological characteristics, and specifications of different materials, eliminating the need for frequent manual adjustments. This adapts to the detection needs of various materials and specifications of barbed sutures, reducing equipment investment costs. Through synchronous pulse triggering of the main encoder and a layered detection process design, multi-parameter synchronous detection is achieved. The detection cycle for a single product is controllable, and the entire process of data acquisition, processing, judgment, and correction is automated, reducing manual intervention and improving production detection efficiency.

[0056] 2. This invention achieves precise tracing of defect root causes by strongly correlating standardized markers of defect location, type, and level with detection parameters, providing data support for front-end production process optimization and forming a closed loop of detection-correction-improvement. Through defect-parameter correlation correction, dynamic iterative optimization of detection parameters is achieved, adapting to fluctuations in the production environment and changes in processes, avoiding batch detection deviations caused by static parameters, and ensuring long-term stability and consistency of detection. Attached Figure Description

[0057] The invention will now be further described with reference to the accompanying drawings.

[0058] Figure 1 This is one of the flowcharts of a multi-parameter visual online detection method for barbed sutures provided in an embodiment of the present invention;

[0059] Figure 2 This is the second flowchart of the multi-parameter visual online detection method for barbed sutures provided in this embodiment of the invention;

[0060] Figure 3 This is a schematic diagram of a multi-parameter visual online inspection device for barbed sutures provided in an embodiment of the present invention. Detailed Implementation

[0061] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0062] like Figures 1 to 3 As shown, the multi-parameter visual online detection method for barbed sutures provided in this embodiment of the invention includes:

[0063] Collect material property data and specification parameters of barbed sutures, formulate suture detection parameters based on target detection accuracy requirements, and use imaging calibration methods to obtain parameter adjustment data based on material reflectivity and morphological characteristics analysis.

[0064] The steps for determining suture inspection parameters include:

[0065] The optical, physical, and processing compatibility characteristics of the suture are collected as material property data, and the suture body parameters and barb parameters are collected as specification parameters.

[0066] Optical properties: material reflectivity, such as high reflectivity of nylon, medium reflectivity of polylactic acid, and low reflectivity of silk; transparency; surface texture density; and uniformity of reflectivity distribution.

[0067] Physical properties: thermal deformation coefficient, such as the change in wire diameter corresponding to a ±1℃ fluctuation in ambient temperature. Surface hardness, whether it is prone to scratches / impacts. Material stability, such as the surface degradation characteristics of absorbable sutures.

[0068] Processing adaptability characteristics: The barb edge features left by die-cutting / laser engraving processes, such as burr residue, edge sharpness, and line bending recovery ability, affect the morphological inspection benchmark.

[0069] The parameters of the line body can include: wire diameter, roundness of the line body cross section, and length specification of the line body.

[0070] The barb parameters can include: barb height, barb angle, barb spacing, and number of barbs.

[0071] Accuracy thresholds are established based on the geometric parameters, defects, and morphology of the object being inspected, and inspection efficiency requirements are determined in conjunction with production speed.

[0072] The accuracy requirements for geometric parameters are as follows: wire diameter measurement accuracy ≤ ±0.002mm, barb height / angle measurement accuracy ≤ ±0.005mm, and barb spacing measurement accuracy ≤ ±0.01mm.

[0073] The defect detection accuracy requirements are as follows: the smallest defect size that can be identified, such as a 0.02mm white tip or a 0.03mm barb; defect positioning accuracy ≤ ±0.1mm; false negative rate ≤ 0.5%; and false positive rate ≤ 1%.

[0074] The accuracy requirements for morphological detection are as follows: accuracy of needle body deformation judgment, allowable bending offset ≤0.03mm; accuracy of needle tail tilting judgment, allowable offset angle ≤2°.

[0075] The required testing efficiency is obtained by calculating the amount of testing per unit time based on the production line speed.

[0076] Based on the detection efficiency requirements, the resolution, shooting frequency, field of view and focal length, light source type and light intensity are determined to obtain the image acquisition parameters.

[0077] Resolution selection: Based on the minimum detection size, if a 0.02mm defect is to be identified, the camera's single pixel accuracy should be ≤0.004mm / pixel. Combined with the field of view, a camera with more than 5 megapixels should be selected.

[0078] Shooting frequency setting: Calculated based on production line speed and minimum detection accuracy. For example, if the wire diameter is 0.1mm, it is required to sample once every 0.05mm. The production line speed is 100m / min = 1.67m / s. The shooting frequency is 1.67m / s ÷ 0.00005m / frame = 33400 frames / second. Taking into account the camera performance, it is rounded to 30000 frames / second.

[0079] Field of view and focal length: Divided by detection area, tip detection uses a small field of view of 11mm×8mm, paired with a telecentric lens. Needle body detection uses a large field of view of 80mm×66mm, paired with a 16mm lens.

[0080] Light source type selection: High-reflective materials use a low-angle ring light source + polarizer. Transparent materials use a coaxial light source. Low-reflective materials use a white parallel backlight.

[0081] Light intensity setting: Adjust according to the reflectivity of the material. Set the intensity to 30%-50% for highly reflective materials and 70%-90% for low-reflective materials to ensure that the image grayscale contrast is ≥30%.

[0082] Based on the accuracy threshold and target defect characteristics, noise reduction parameters, edge enhancement parameters, and illumination correction parameters are selected as processing parameters. Workstations are divided according to the preset detection logic, and the camera trigger interval is set based on the pulse of the main encoder of the production line to obtain the detection process parameters.

[0083] Based on the historical testing parameters and results of similar products, correction logic and process characteristic correction parameters are extracted. The image acquisition parameters, processing parameters, and testing process parameters are adjusted to obtain the suture detection parameters.

[0084] If historical data shows that a certain material still exhibits low contrast even at 70% illumination intensity, then the current light source intensity will be adjusted to 80% and supplemented with localized lighting. If historical algorithm thresholds cause minor defects such as barbs to be missed, then the defect area determination threshold will be lowered.

[0085] If the manufacturing process involves laser engraving of barbs, the edge enhancement threshold can be appropriately reduced, such as a Canny low threshold of 40. If the barbs are die-cut, a parameter for distinguishing burrs from defects should be added, such as setting a burr width threshold ≤ 0.01mm to be considered normal.

[0086] If the temperature and humidity fluctuate significantly during the production process, a thermal deformation compensation coefficient is added to the algorithm parameters. For example, for every 1°C increase in temperature, the wire diameter qualification threshold increases by 0.001 mm.

[0087] The steps for analyzing and obtaining parameter adjustment data include:

[0088] The average reflectivity, reflectance distribution uniformity, and grayscale difference between the suture and the background are extracted as material reflectance characteristics for the barbed suture.

[0089] Average reflectance was measured using a reflectance tester. Under actual lighting conditions at the testing station, 10 sampling points were evenly selected on the suture surface, covering the suture body, barb tip, and barb root. The reflectance of each point was measured (range 0%-100%), and the average value R was calculated. The sutures were then classified according to the following standards:

[0090] High reflectivity: R≥60%, such as nylon. Medium reflectivity: 30%<R<60%, such as polylactic acid. Low reflectivity: R≤30%, such as silk.

[0091] Evenness of reflectance distribution (U): Calculate the standard deviation σ of reflectance at 10 sampling points, defined as U = σ / R × 100%, reflecting the presence of localized bright spots / shadows in the reflection.

[0092] Uniform: U≤10%. Moderately uniform: 10%<U<20%. Non-uniform: U≥20%.

[0093] Indicator 3: Defect-Background Gray-Level Difference (ΔG) Images of standard suture samples are collected, and the average gray-level values ​​of the defect area and the surrounding background are extracted using image processing software. The difference ΔG is then calculated.

[0094] High contrast ratio: ΔG ≥ 40. Medium contrast ratio: 20 < ΔG < 40. Low contrast ratio: ΔG ≤ 20, such as minor scratches on low-reflectivity materials.

[0095] With the goal of clearly extracting image characteristics, it determines whether the reflective properties of the current material meet the detection requirements and identifies the type of deviation to obtain the reflective property deviation.

[0096] Deviation Type 1: Excessive Reflection (R≥60% and U≥15%) Manifestations: Localized light spots appear in the image, obscuring the barb edges and causing edge detection failure. Impact: Barb height and angle measurement errors exceed ±0.01mm, increasing the missed detection rate of minor defects.

[0097] Deviation Type 2: Insufficient Reflection (R≤30% and ΔG≤20) Characteristics: The overall image is dark, defects are not clearly distinguishable from the background, and the outline of barbs is blurred. Impact: The recognition rate of low-contrast defects such as surface scratches and barb defects is less than 70%.

[0098] Deviation Type 3: Uneven Reflection Distribution (U≥20%) Manifestations: Overexposure in some areas, underexposure in others, and conflicting compatibility of detection parameters at different locations along the same suture line. Impact: Local deviations occur in line diameter measurement, such as misjudging the line diameter as too small in the spot area.

[0099] The sharpness of the barb edge, the consistency of the line shape, the uniformity of the barb distribution, and the density of the surface texture of the barb suture are extracted as morphological features to determine whether they affect the deviation of the morphological features obtained from parameter measurement and defect identification.

[0100] Based on the deviations in reflective properties and morphological characteristics, the adjustment direction and parameters of the corresponding matching suture detection parameters are determined.

[0101] The steps to obtain morphological features may include: Barb edge sharpness (S): acquiring a microscopic image of the barb tip, calculating the gray-scale change rate of the barb edge using an edge gradient algorithm, and taking the average gradient S of 10 barbs.

[0102] Sharp: S≥80. Medium sharp: 40<S<80. Blunt round: S≤40, such as burrs remaining from the die-cutting process or wear of barbs due to material aging.

[0103] Consistency of suture shape (C): A single suture segment is scanned using a laser profilometer to extract the bending offset of the suture's central axis, and the ratio C of the maximum offset to the suture diameter is calculated.

[0104] Consistency: C≤5%. Minor deviation: 5%<C<10%. Severe deviation: C≥10%, such as bending of the production line caused by uneven production traction.

[0105] Barb distribution uniformity (D): Measure the spacing between all barbs within a single suture segment and calculate the ratio D of the standard deviation of the spacing to the mean spacing.

[0106] Uniform: D≤8%. Moderate uniformity: 8%<D<15%. Non-uniform: D≥15%, such as the deviation in barb spacing caused by die-cutting blade wear.

[0107] Surface texture density (T): The percentage of texture pixels per unit area, T, is calculated by binarizing the surface image of the suture line.

[0108] Low texture: T≤10%, such as smooth polylactic acid yarn. Medium texture: 10%<T<30%. High texture: T≥30%, such as braided nylon yarn.

[0109] Deviation type 1: Blunt rounded barb edge (S≤40) Effect: Edge positioning deviation during barb height and angle measurement, accuracy drops from ±0.005mm to more than ±0.01mm.

[0110] Deviation Type 2: Inconsistent line shape (C≥10%) Impact: The measurement coordinate system reference is offset, and the measurement reference for line diameter and barb spacing is incorrect, resulting in systematic deviation.

[0111] Deviation type 3: High texture interference (T≥30%) Impact: Texture is confused with surface defects, and the false detection rate increases by more than 15%.

[0112] Based on the preset accuracy requirements and adjustment direction, the adjustment coefficients of the adjustment parameters are calculated to obtain the parameter adjustment data.

[0113]

[0114] The target detection parameters are obtained by adjusting the suture detection parameters based on the parameter adjustment data. Multi-view image data of the current barbed suture is collected and processed using a multi-view detection method to obtain multi-dimensional detection data.

[0115] A mapping table for adjusting suture detection parameters and parameter adjustment data is established by mapping the parameters one-to-one.

[0116] Adjust the parameter mapping table:

[0117]

[0118] Based on the directness of the impact of different adjustment parameters on the detection effect and the response speed, the parameter adjustment priority is obtained by prioritizing them.

[0119] Step 1: Adjust the priority 1 parameter. For example, replace the original light source intensity of 12V with 5.3V in the adjustment data, correct the original edge detection low threshold of 50 to 25 (50×0.5), and replace the original shape correction coefficient of 1.0 with 0.87.

[0120] Step 2: Adjust the priority 2 parameter. For example, replace the original exposure time of 50μs with 40μs, turn on the polarizer, and switch the sub-pixel interpolation mode from bilinear to cubic interpolation.

[0121] Step 3: Adjust the priority 3 parameters. For example, adjust the original texture filter intensity from 0 to 0.8, and adjust the original minimum defect area threshold from 0.0004mm² to 0.0003mm², while keeping the trigger interval unchanged.

[0122] Verification Logic 1: After parameter adjustment, the direct accuracy requirement must be met. For example, after adjusting the light source intensity and edge threshold, collect the measured height values ​​of 10 barbs and compare them with the standard value. If the error is ≤ ±0.005mm, it is qualified. If the error exceeds the standard, make a fine adjustment by multiplying the adjustment range by 1.1, such as adjusting the light source intensity from 5.3V to 5.8V.

[0123] Verification Logic 2: Parameter adjustments should not affect accuracy. For example, after adjusting the texture filter intensity, it should be confirmed that the barb contour has not been over-filtered and the defect recognition rate has not decreased. If it has decreased, the filter intensity should be reduced.

[0124] If increasing the light source intensity and decreasing the exposure time at the same time results in an image that is too bright or too dark, the image grayscale contrast ΔG needs to be measured again. If ΔG < 30, the light source intensity should be increased by 10% or the exposure time should be increased by 5% until ΔG ≥ 30.

[0125] The adjustment operation is performed step by step according to the parameter adjustment priority and the adjustment parameter mapping table. With the target detection accuracy requirement as the constraint, the adjustment range of each adjustment parameter is checked to see if it meets the standard. If it does, the target detection parameter is output; otherwise, the adjustment range is corrected.

[0126] Step 1: Adjust the priority 1 parameter. For example, replace the original light source intensity of 12V with 5.3V in the adjustment data, correct the original edge detection low threshold of 50 to 25 (50×0.5), and replace the original shape correction coefficient of 1.0 with 0.87.

[0127] Step 2: Adjust the priority 2 parameter. For example, replace the original exposure time of 50μs with 40μs, turn on the polarizer, and switch the sub-pixel interpolation mode from bilinear to cubic interpolation.

[0128] Step 3: Adjust the priority 3 parameters. For example, adjust the original texture filter intensity from 0 to 0.8, and adjust the original minimum defect area threshold from 0.0004mm² to 0.0003mm², while keeping the trigger interval unchanged.

[0129] Verification Logic 1: After parameter adjustment, the direct accuracy requirement must be met. For example, after adjusting the light source intensity and edge threshold, collect the measured height values ​​of 10 barbs and compare them with the standard value. If the error is ≤ ±0.005mm, it is qualified. If the error exceeds the standard, make a fine adjustment by multiplying the adjustment range by 1.1, such as adjusting the light source intensity from 5.3V to 5.8V.

[0130] Verification Logic 2: Parameter adjustments should not affect accuracy. For example, after adjusting the texture filter intensity, it should be confirmed that the barb contour has not been over-filtered and the defect recognition rate has not decreased. If it has decreased, the filter intensity should be reduced.

[0131] If increasing the light source intensity and decreasing the exposure time at the same time results in an image that is too bright or too dark, the image grayscale contrast ΔG needs to be measured again. If ΔG < 30, the light source intensity should be increased by 10% or the exposure time should be increased by 5% until ΔG ≥ 30.

[0132] Based on the target detection parameters, determine whether the multi-dimensional detection data meets the preset threshold. If yes, output the qualified judgment result; otherwise, mark the defect location type.

[0133] The steps to determine whether multi-dimensional detection data meets the preset threshold include:

[0134] Based on the target detection parameters, the judgment standard thresholds are set in layers according to parameter type, detection area and defect level.

[0135] The judgment criteria threshold may include: wire diameter threshold: qualified range = target wire diameter ± 0.002mm, such as 0.1mm wire diameter corresponds to 0.098mm-0.102mm.

[0136] Barb parameter thresholds: height ≥ 0.075mm, angle 38°-52°, spacing 0.29mm-0.31mm, number ≥ 23 / 30mm.

[0137] Needle body / needle tail morphology thresholds: deformation ≤ 0.03 mm, needle tail tilt angle ≤ 2°, needle tip curvature radius ≤ 0.03 mm.

[0138] Defect size thresholds: Critical defects, area ≥ 0.001 mm²; Serious defects, 0.0005 mm² - 0.001 mm²; Minor defects, < 0.0005 mm².

[0139] Defect density threshold: ≤2 severe defects, ≤3 minor defects, and no fatal defects within a single suture segment (50mm).

[0140] Acceptance criteria: All geometric parameters are within the acceptable threshold + No fatal defects + Number of serious defects ≤ threshold + Number of minor defects ≤ threshold.

[0141] Non-compliance triggering rules: If any geometric parameter exceeds the threshold, or a fatal defect exists, or the number of serious / minor defects exceeds the limit, the result is a veto.

[0142] The multi-dimensional detection data are compared with the target detection data one by one according to the judgment standard threshold, and the qualified judgment result is output.

[0143] Wire diameter verification: Calculate the average wire diameter of multiple sections. If it is within ±0.002mm of the target wire diameter, it is qualified. If it exceeds the range, it is directly judged as unqualified. Record the wire diameter unevenness / out-of-tolerance preliminary defect.

[0144] Barb parameter verification: Check the height, angle and spacing of each barb one by one, and take the mean and standard deviation of all barb measurements. If the mean exceeds the threshold or the proportion of a single barb parameter exceeding the threshold is greater than 5%, the barb parameter is judged to be abnormal.

[0145] Needle body / needle tail morphology check: If the deformation is greater than 0.03mm, the tilt angle is greater than 2°, or the radius of curvature of the needle tip is greater than 0.03mm, the morphology is deemed unqualified.

[0146] The steps for marking the location type of defects include:

[0147] Defect coordinates are determined based on multi-dimensional detection data. Combined with the divided detection areas, the location of the defect is marked, and the corresponding camera view image is associated to mark the defect image area.

[0148] Defect image regions are classified according to geometric parameters, surface defects, and morphology to determine the defect image type.

[0149] Geometric parameters can include wire diameter deviation, insufficient barb height, needle deformation, etc.; surface defects can include white tip, barb defects, needle body pitting, nodules, etc.; morphology can include needle tail curling side, barb angle deviation, etc.

[0150] The defect image level is labeled according to the defect size and impact of the defect image area, and the defect location type is obtained by classifying it according to product number-non-conformance-time stamp.

[0151] By analyzing the correlation between defect location type and suture detection parameters, defect detection correlation correction data is obtained, and the target detection parameters are adjusted to obtain accurate detection parameters.

[0152] The steps to obtain defect detection correlation correction data include:

[0153] Standardized defect location types are processed to obtain standard defect data. The suture detection parameters of the corresponding defective products are extracted and matched with the standard defect data according to the product number and detection timestamp.

[0154] The association analysis dimensions are obtained by associating defect type with parameter type, defect location with parameter effective range, and defect index with parameter value deviation.

[0155] The defect type is related to the parameter type, such as whether the barb defect is strongly correlated with the barb measurement-related parameters.

[0156] The location of the defect is related to the effective range of the parameter, such as whether the defect in the needle tip area is only related to the parameters of the needle tip detection station.

[0157] Defect indicators are related to parameter value deviations. For example, the greater the deviation in barb height, the lower the corresponding edge detection threshold.

[0158] Based on the correlation analysis dimensions, parameter value distribution characteristics, defect occurrence rate correlation characteristics, and indicator parameter correlation characteristics are extracted from standard defect data as correlation features.

[0159] Parameter value distribution characteristics: Grouped by defect type, the distribution of detection parameter values ​​corresponding to each group of defects is statistically analyzed and compared with the parameter value distribution of qualified products to calculate the deviation rate. For example, the average light source intensity corresponding to the barb defect is 5.2V, while the average value of qualified products is 5.8V, with a deviation rate of 10.3%.

[0160] Defect incidence rate correlation characteristics: Calculate the defect incidence rate for different parameter value ranges. For example, when the light source intensity is 5.0-5.5V, the burr defect incidence rate is 12%. When it is 5.5-6.0V, the incidence rate is 3%. Screen out the parameter threshold ranges where the defect incidence rate changes abruptly.

[0161] Correlation between defect quantification indicators and parameters: The correlation between defect quantification indicators and detection parameter values ​​is calculated using the Pearson correlation coefficient. |r|≥0.6 indicates a strong correlation, and 0.3≤|r|<0.6 indicates a moderate correlation. For example, the correlation between barb height deviation and edge detection threshold is r=-0.72, which is a negative correlation. The lower the threshold, the greater the deviation.

[0162] Sensitivity is calculated using the defect improvement rate as a sensitivity metric. This quantifies the impact of each suture detection parameter on associated features, and parameters reaching a preset sensitivity threshold are selected as high-sensitivity parameters. The sensitivity calculation formula is expressed as follows:

[0163]

[0164] In the formula, It's about sensitivity. It is the defect incidence rate after parameter adjustment. It is the defect incidence rate before adjustment.

[0165] Based on highly sensitive parameters and associated features, parameter correction rules are formulated to obtain defect detection associated correction data.

[0166] Related Correction Table:

[0167]

[0168] The steps to adjust and obtain accurate detection parameters include:

[0169] Based on the target detection parameters, a one-to-one correspondence is established with the defect detection-related correction data to form a two-way mapping relationship. Scenario correction data is obtained by filtering according to the product material, specifications, and defect type currently being detected.

[0170] Bidirectional mapping table:

[0171]

[0172] Correction priorities are determined based on parameter adjustment priorities and the degree of impact, and the adjustment parameter values ​​corresponding to the correction priorities are calculated based on scenario correction data.

[0173] Priority 1 for correction: High-sensitivity parameters corresponding to fatal / critical defects, such as the edge detection threshold associated with barb defects and the light source intensity associated with pinhead whiteheads, can be adjusted to significantly reduce the critical defect rate.

[0174] Correction priority 2: Medium sensitivity parameters corresponding to minor defects, such as the texture filtering intensity associated with pinholes.

[0175] Correction priority 3: Does not directly affect defect detection, but needs to be adapted to the auxiliary parameters adjusted by priority 1-2, such as adjusting the exposure time after adjusting the light source intensity to avoid image over-excitation.

[0176] Priority 1 parameter adjustment: Calculated directly based on the correction direction and magnitude of the correction data, as shown in the example below:

[0177] Original target parameters: edge detection low threshold = 22, correction requirement +10%, after adjustment = 22 × (1 + 10%) = 24.2.

[0178] Original target parameters: light source intensity = 5.8V, correction requirement +8%, after adjustment = 5.8 × (1 + 8%) ≈ 6.26V.

[0179] Priority 2 parameter adjustment: After calculation based on the adjustment magnitude, compatibility with other parameters needs to be verified. Example:

[0180] Original target parameters: texture filter intensity = 0.6, correction requirement +20%, adjusted = 0.72, verified to not affect the retention of the extracted barb contour.

[0181] Correction priority 3 parameter adjustment: Assisted adjustment based on parameter linkage relationships, example:

[0182] When the light source intensity increases from 5.8V to 6.3V, in order to avoid overexposure of the image, the exposure time is adjusted from 44μs to 44×(1-4%)≈42.2μs according to the linkage rule that the exposure time decreases by 5% for every 10% increase in light source intensity.

[0183] Based on the parameter detection results and equipment boundaries, a conflict coordination strategy is formulated, and the target detection parameters are adjusted according to the adjusted parameter values ​​to obtain accurate detection parameters.

[0184] Conflict coordination strategies may include:

[0185] Conflicts in multiple corrections for the same parameter: take the correction data with higher confidence and better expected improvement rate. If the confidence levels are close, take the average of the correction magnitudes.

[0186] Cross-parameter linkage conflict: A primary-secondary balance principle is adopted, prioritizing the effectiveness of high-priority parameters and fine-tuning the compensation of low-priority parameters. Example:

[0187] Increasing the edge detection threshold raises the rate of missed detection for minor defects. However, it is possible to simultaneously lower the defect area determination threshold to balance the risk of missed detection.

[0188] Device boundary conflict: Set parameters according to the upper / lower limit of the device, and supplement with other parameters to assist in correction. Example:

[0189] The upper limit of the light source intensity is 8V, and the correction requirement is +50% to 8.7V. First, set it to 8V, and then extend the exposure time by 10% to compensate for insufficient brightness.

[0190] Precise detection parameters may include: Light source intensity: The original target detection parameter value is 5.8V. Based on the correlation correction data corresponding to the needle tip white head defect (confidence level 93%, expected defect improvement rate 75%), it needs to be increased by 8%. The adjusted precise detection parameter value is 6.3V, and the effective range is nylon barbed sutures with a wire diameter of 0.1-0.3mm.

[0191] Exposure time: The original target detection parameter value was 44μs. Due to the need for compensation after adjusting the light source intensity to avoid image overexposure, the value was reduced by 4% according to the linkage rule. The adjusted accurate detection parameter value is 42μs.

[0192] Polarizer status: No associated correction data, accurate detection parameter values ​​maintain the original target detection parameter on state.

[0193] Edge detection low threshold: The original target detection parameter value is 22. Based on the correlation correction data corresponding to the barb defect (confidence level 96%, expected defect improvement rate 66.7%), it needs to be increased by 10%. The adjusted accurate detection parameter value is 24, which is effective for nylon barb sutures with a diameter of 0.1-0.3mm.

[0194] Subpixel interpolation mode: No associated correction data, accurate detection parameter values ​​maintain the original target detection parameters in a fifth-order interpolation mode.

[0195] Morphological correction coefficient: The original target detection parameter value is 0.87. Based on the correlation correction data corresponding to the barb defect, it needs to be increased by 5%. The adjusted accurate detection parameter value is 0.91, and the effective range is nylon barb sutures with a diameter of 0.1-0.3mm.

[0196] Minimum defect area threshold: The original target detection parameter value was 0.0003 mm². Due to the need to compensate for the risk of missing small defects after increasing the edge detection threshold, it was reduced by 16.7%, and the adjusted accurate detection parameter value is 0.00025 mm².

[0197] Based on the same general inventive concept, this invention also protects a multi-parameter visual online inspection device suitable for barbed sutures, comprising:

[0198] The intelligent parameter optimization module is used to collect material property data and specification parameters of the barbed suture, formulate suture detection parameters in combination with the target detection accuracy requirements, and obtain parameter adjustment data by analyzing the material's reflective properties and morphological characteristics using an imaging calibration method.

[0199] The online image detection module is used to adjust the suture detection parameters according to the parameter adjustment data to obtain the target detection parameters, collect multi-view image data of the current barbed suture, and process it using a multi-view detection method to obtain multi-dimensional detection data.

[0200] The defect determination and marking module is used to determine whether the multi-dimensional detection data meets the preset threshold based on the target detection parameters. If it does, the module outputs a qualified judgment result; otherwise, it marks the defect location type.

[0201] The parameter closed-loop correction module is used to analyze the correlation between defect location type and suture detection parameters to obtain defect detection correlation correction data, and adjust the target detection parameters to obtain accurate detection parameters.

[0202] In summary, the multi-parameter visual online inspection method and equipment for barbed sutures provided in this embodiment effectively solves the problem of fluctuating detection accuracy for sutures of different materials and batches through adaptive calibration based on material characteristics and intelligent parameter optimization, significantly improving the detection accuracy. Employing a multi-view camera array and region segmentation strategy, the three-dimensional structural features of the suture are covered from multiple angles, eliminating blind spots in single-view detection and improving defect identification capabilities. Automated parameter adjustment and closed-loop optimization mechanisms significantly reduce the workload of manually adjusting detection parameters and lower the technical requirements for operators. Through defect detection correlation correction and parameter closed-loop optimization, the system can automatically adjust detection parameters based on historical defect data, possessing continuous learning and optimization capabilities. Complete traceability of detection data is achieved, facilitating quality analysis and problem localization.

[0203] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0204] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-parameter visual online detection method for barbed sutures, comprising: characterized in that: Collect material property data and specification parameters of barbed sutures, formulate suture detection parameters in combination with target detection accuracy requirements, and use imaging calibration methods to obtain parameter adjustment data based on material reflectivity and morphological characteristics analysis; The target detection parameters are obtained by adjusting the suture detection parameters according to the parameter adjustment data. Multi-view image data of the current barbed suture is collected and processed using a multi-view detection method to obtain multi-dimensional detection data. Based on the target detection parameters, determine whether the multi-dimensional detection data meets the preset threshold. If yes, output the qualified judgment result; otherwise, mark the defect location type. By analyzing the correlation between the defect location type and the suture detection parameters, defect detection correlation correction data is obtained, and the target detection parameters are adjusted to obtain accurate detection parameters.

2. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The steps for determining the suture detection parameters include: The optical, physical, and processing compatibility properties of the suture are collected as material property data, and the suture body parameters and barb parameters are collected as specification parameters. Accuracy thresholds are determined based on the geometric parameters, defects, and morphology of the object being inspected, and inspection efficiency requirements are determined in conjunction with production speed. Based on the detection efficiency requirements, the resolution, shooting frequency, field of view and focal length, light source type and light intensity are determined to obtain the image acquisition parameters. Based on the accuracy threshold and target defect characteristics, noise reduction parameters, edge enhancement parameters and illumination correction parameters are selected as processing parameters. Workstations are divided according to the preset detection logic. The camera trigger interval is set based on the pulse of the main encoder of the production line to obtain the detection process parameters. Based on the historical detection parameters and effects of similar products, the correction logic and process characteristic correction parameters are extracted, and the image acquisition parameters, the processing parameters, and the detection process parameters are adjusted to obtain the suture detection parameters.

3. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The steps for analyzing and obtaining the parameter adjustment data include: The average reflectivity, reflectance distribution uniformity, and defect-background grayscale difference of the material optical properties of the barbed suture are extracted as the reflectance characteristics of the material. With the goal of clearly extracting image characteristics, it determines whether the reflective properties of the current material meet the detection requirements and identifies the type of deviation to obtain reflective property deviation. The sharpness of the barb edge, the consistency of the line shape, the uniformity of the barb distribution, and the density of the surface texture of the barb suture are extracted as the morphological features, and it is determined whether they affect the deviation of the morphological features obtained by parameter measurement and defect identification. Based on the reflective characteristic deviation and the morphological feature deviation, the adjustment direction and adjustment parameters of the suture detection parameters are matched accordingly; Based on the preset accuracy requirements and the adjustment direction, the adjustment coefficient of the adjustment parameter is calculated to obtain the parameter adjustment data.

4. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The steps for adjusting the target detection parameters include: Establish a mapping table for adjustment parameters by mapping the suture detection parameters one-to-one with the parameter adjustment data; Based on the directness of the impact of different adjustment parameters on the detection effect and the response speed, the parameter adjustment priority is obtained by prioritizing them. The adjustment operation is performed step by step according to the parameter adjustment priority and the adjustment parameter mapping table. With the target detection accuracy requirement as a constraint, the adjustment range of each adjustment parameter is checked to see if it meets the standard. If it does, the target detection parameter is output; otherwise, the adjustment range is corrected.

5. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The steps to obtain the multi-dimensional detection data include: The multi-view camera array is calibrated according to the target detection parameters, the camera acquisition interval is set, the detection area is divided according to the suture structure, the image is acquired, and the multi-view image data is obtained by storing the data according to the product number-detection area-camera number-time stamp. The multi-view image data is subjected to grayscale conversion, noise reduction, and illumination correction, and geometric correction and region cropping are performed to obtain preliminary processed image data; Feature extraction is performed on the pre-processed image data to obtain contour defect features, and geometric dimensions, morphological accuracy, and feature surface defect parameters are calculated to obtain the multi-dimensional detection data.

6. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The steps for determining whether the multi-dimensional detection data meets the preset threshold include: Based on the target detection parameters, judgment standard thresholds are set in layers according to parameter type, detection area, and defect level. The multi-dimensional detection data is compared with the target detection data one by one according to the judgment standard threshold, and the qualified judgment result is output.

7. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The step of marking the defect location type includes: Based on the multi-dimensional detection data, the defect coordinates are determined, and combined with the divided detection areas, the location of the defect is marked, the corresponding camera view image is associated, and the defect image area is marked. The defect image regions are classified according to geometric parameters, surface defects, and morphology to determine the defect image type; The defect image level is labeled according to the defect size and impact of the defect image area, and the defect location type is obtained by classifying it according to product number-non-conformance-time stamp.

8. The multi-parameter visual online detection method for barbed sutures according to claim 1, characterized in that: The steps for obtaining the defect detection correlation correction data include: The defect location type is standardized to obtain standard defect data. The suture detection parameters of the corresponding defective products are extracted and matched with the standard defect data according to the product number and detection timestamp. The association analysis dimensions are obtained by associating defect type with parameter type, defect location with parameter effective range, and defect index with parameter value deviation. Based on the aforementioned correlation analysis dimensions, parameter value distribution characteristics, defect occurrence rate correlation characteristics, and indicator parameter correlation characteristics are extracted from the standard defect data as correlation features. Sensitivity is calculated using the defect improvement rate as a sensitivity index. The influence of each suture detection parameter on the associated features is quantified, and parameters that reach a preset sensitivity threshold are selected as high-sensitivity parameters. Based on the high-sensitivity parameters and the associated features, parameter correction rules are formulated to obtain the defect detection associated correction data.

9. The multi-parameter visual online detection method for barbed sutures according to claim 8, characterized in that: The steps for adjusting to obtain the precise detection parameters include: Based on the target detection parameters, a one-to-one correspondence is established with the defect detection associated correction data to form a two-way mapping relationship, and the scene correction data is obtained by filtering according to the product material, specifications and defect type currently being detected. Based on the parameter adjustment priority and the degree of impact, a correction priority is determined, and the adjustment parameter value corresponding to the correction priority is calculated according to the scenario correction data. Based on the parameter detection results and equipment boundaries, a conflict coordination strategy is formulated, and the target detection parameters are adjusted according to the adjustment parameter values ​​to obtain the accurate detection parameters.

10. A multi-parameter visual online inspection device for barbed sutures, which is applied to the multi-parameter visual online inspection method for barbed sutures as described in any one of claims 1 to 9, characterized in that, The detection equipment includes: The intelligent parameter optimization module is used to collect material property data and specification parameters of the barbed suture, formulate suture detection parameters in combination with target detection accuracy requirements, and obtain parameter adjustment data by analyzing the material's reflective properties and morphological characteristics using an imaging calibration method. The online image detection module is used to adjust the suture detection parameters according to the parameter adjustment data to obtain the target detection parameters, collect multi-view image data of the current barbed suture, and process it using a multi-view detection method to obtain multi-dimensional detection data; The defect determination and marking module is used to determine whether the multi-dimensional detection data meets the preset threshold based on the target detection parameters. If yes, it outputs a qualified judgment result; otherwise, it marks the defect location type. The parameter closed-loop correction module is used to analyze the correlation between the defect location type and the suture detection parameters to obtain defect detection correlation correction data, and to adjust the target detection parameters to obtain accurate detection parameters.

Images