A surface detection method for shaft parts machining

Abstract

The present application relates to the technical field of surface defect detection, and particularly relates to a surface detection method for shaft part machining, which screens a defect suspected area according to a cross-section diameter parameter obtained in each detection area along a characteristic direction, screens a risk profile according to a distribution coefficient of a texture profile, marks a first risk sub-profile or determines a characteristic risk profile based on a texture representation value of the risk profile, determines a second risk sub-profile according to a characteristic profile point and screens the characteristic risk profile, constructs a defect trend vector according to the risk profile to determine a profile trend representation coefficient, adjusts an illumination parameter for defect detection of the characteristic risk profile, and performs defect detection on the characteristic risk profile again based on the adjusted illumination parameter. The present application realizes rapid identification of a region with a risk of shadow obstruction, adaptively adjusts a detection mode according to actual defect characteristics of shaft parts, and improves the precision and reliability of surface defect detection.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of surface defect detection, in particular to a surface detection method for shaft part machining. BACKGROUND

[0002] Shaft parts are typical parts in mechanical products, which are used for supporting driving parts and transmitting torque, and the machining quality of the shaft parts is directly related to the performance and service life of mechanical equipment. In the machining process of the shaft parts, surface defect detection is a key link. Shaft parts, such as automobile transmission shafts, aero-engine main shafts and machine tool leadscrews, are core components of mechanical transmission systems, and surface defects of the shaft parts directly affect the running accuracy, fatigue life and safety of the equipment. Axial / radial cracks are one of the main causes of shaft fracture. Therefore, high-precision surface defect detection is a key link to ensure the reliability of shaft parts.

[0003] Traditional manual detection methods have many drawbacks, such as low efficiency, susceptibility to subjective factors, poor reliability and the like, and are difficult to meet the needs of modern industrial automated production. Although existing detection technologies are various, each has limitations. Ultrasonic detection can detect internal defects, but has low sensitivity for detecting small surface defects and requires high technical level of operators. Existing visual detection systems often use a single detection algorithm and detection parameters for shaft parts of different shapes and sizes, cannot adaptively adjust the detection parameters according to the actual characteristics of the shaft parts, and cannot determine the optimal illumination angle, resulting in coexistence of over-detection and missed detection and unsatisfactory detection effect. Therefore, improving the surface detection precision and reliability of shaft parts is a technical problem to be solved.

[0004] For example, Chinese Patent No. CN118392875B discloses a shaft part surface nondestructive detection system and method, which comprises a target confirmation unit, a quantitative detection unit and a terminal display unit, and further comprises an image acquisition unit. The image acquisition unit is used for setting multiple CCD cameras in different directions of a target detection area. The target shaft part is collected at multiple angles through the multiple CCD cameras, so that multiple target shaft part images at different angles are obtained, and the multiple target shaft part images at different angles are transmitted to an image preprocessing unit. The application sets multiple CCD cameras in different directions through the image acquisition unit, can collect the target shaft part at multiple angles, and reduces errors through the image preprocessing unit for denoising and enhancing the collected images.

[0005] The prior art also has the following problems:

[0006] The prior art does not consider that the shaft part is prone to have a shadow blocking area due to its unique structural characteristics, which affects the detection accuracy of surface defects, and the prior art cannot quickly identify the area at risk of shadow blocking, and cannot adaptively adjust the detection method according to the actual defect characteristics of the shaft part, thereby affecting the accuracy and reliability of surface defect detection. SUMMARY

[0007] Therefore, the present application provides a surface detection method for shaft part machining to overcome the problems in the prior art that the area at risk of shadow blocking cannot be quickly identified, and the detection method cannot be adaptively adjusted according to the actual defect characteristics of the shaft part, thereby affecting the accuracy and reliability of surface defect detection.

[0008] To achieve the above-mentioned purpose, the present application provides a surface detection method for shaft part machining, comprising:

[0009] A surface image of the shaft part to be detected is obtained at a preset illumination angle and contour recognition is performed, the shaft part is divided into a plurality of detection areas, a plurality of cross-sectional diameter parameters are obtained in each detection area along a characteristic direction, cross-sectional fluctuation representation parameters are determined based on the cross-sectional diameter parameters to screen defect suspected areas;

[0010] The characteristic direction is determined according to the preset illumination angle;

[0011] A surface image of the defect suspected area is obtained and a texture contour is recognized, a risk contour is screened according to a distribution coefficient of the texture contour, and a first risk sub-contour or a characteristic risk contour is marked based on a texture representation value of the risk contour;

[0012] A surface image of an adjacent detection area of the defect suspected area is obtained based on a characteristic contour point, and a risk contour is recognized, the surface image contains the characteristic contour point, a second risk sub-contour is determined according to the characteristic contour point, and a characteristic risk contour is screened;

[0013] The characteristic contour point is determined according to the first risk sub-contour;

[0014] A defect trend vector is constructed according to the risk contour to determine a contour trend representation coefficient, and an illumination parameter for defect detection of the characteristic risk contour is adjusted based on the contour trend representation coefficient;

[0015] The characteristic risk contour is again subjected to defect detection based on the adjusted illumination parameter.

[0016] Further, determining the cross-sectional fluctuation representation parameter comprises,

[0017] The cross-section diameter parameters of each preset detection point in the detection area are sorted along the characteristic direction, the difference between the cross-section diameter parameter of a previous preset detection point and the cross-section diameter parameter of a subsequent preset detection point in adjacent preset detection points is calculated, and the difference is determined as the cross-section fluctuation characteristic parameter.

[0018] Further, the process of screening the defect suspected area includes,

[0019] If the cross-section fluctuation characteristic parameter in the detection area meets the defect risk determination condition, the detection area is screened as a defect suspected area.

[0020] The defect risk determination condition is that at least one cross-section fluctuation characteristic parameter in the detection area exceeds a preset cross-section fluctuation characteristic threshold.

[0021] Further, the process of screening the risk profile includes,

[0022] The interval distance between each preset point on the groove profile and an adjacent groove profile is obtained, and the interval distance variance is determined as the distribution coefficient of the groove profile.

[0023] If the distribution coefficient of the groove profile meets the risk profile determination condition, the groove profile is screened as a risk profile.

[0024] The risk profile determination condition is that the distribution coefficient exceeds a preset distribution coefficient reference value.

[0025] Further, the process of marking the first risk sub-profile or determining the characteristic risk profile includes,

[0026] The interval distance between the end points of the risk profile is determined as the groove characteristic value of the risk profile.

[0027] If the groove characteristic value of the risk profile meets the first characteristic risk profile determination condition, the risk profile is determined as a characteristic risk profile.

[0028] If the groove characteristic value of the risk profile does not meet the first characteristic risk profile determination condition and there is a characteristic groove end point, the risk profile is marked as a first risk sub-profile.

[0029] The first characteristic risk profile determination condition is that the groove characteristic value exceeds a preset groove characteristic value threshold.

[0030] Further, the process of determining the characteristic profile point includes,

[0031] If the end point of the first risk sub-profile is on the profile edge of the surface image, the end point is determined as a characteristic profile point.

[0032] Further, the process of determining the second risk sub-profile and screening the feature risk profile comprises,

[0033] marking the risk profile where the feature profile point is located as a second risk sub-profile;

[0034] if the first risk sub-profile and the second risk sub-profile meet a second feature risk profile determination condition, determining the first risk sub-profile and the second risk sub-profile as a feature risk profile together;

[0035] wherein the second feature risk profile determination condition is that the sum of the texture representation value of the first risk sub-profile and the texture representation value of the second risk sub-profile exceeds a preset texture representation value threshold.

[0036] Further, the process of determining the profile trend representation coefficient comprises,

[0037] taking any endpoint of the feature risk profile as the starting point of a vector, calculating the interval distance between the remaining endpoints on the feature risk profile and the endpoint, and taking the endpoint with the maximum interval distance as the terminal point of the vector to construct the defect trend vector;

[0038] respectively taking the component vector of the defect trend vector in the direction parallel to the axial direction of the shaft part as a first component vector, and taking the component vector of the defect trend vector in the direction perpendicular to the axial direction of the shaft part as a second component vector;

[0039] determining the ratio of the module length of the first component vector to the module length of the second component vector as the profile trend representation coefficient.

[0040] Further, the process of adjusting the illumination parameter for defect detection of the feature risk profile comprises,

[0041] if the profile trend representation coefficient of the feature risk profile exceeds a preset profile trend representation coefficient threshold range, setting the illumination direction for defect detection of the feature risk profile along the radial direction of the shaft part, and the increase amount of the illumination angle is in a positive correlation with the profile trend representation coefficient;

[0042] if the profile trend representation coefficient of the feature risk profile is less than a preset profile trend representation coefficient threshold range, setting the illumination direction for defect detection of the feature risk profile along the axial direction of the shaft part, and the increase amount of the illumination angle is in a positive correlation with the profile trend representation coefficient.

[0043] Further, the illumination angle is the included angle between the light source and the tangent plane of the feature risk profile.

[0044] Compared with the prior art, the present application has the beneficial effects that the present application obtains the surface image of the shaft part to be detected at a preset illumination angle and performs contour recognition, divides the shaft part into a plurality of detection regions, obtains a plurality of cross-sectional diameter parameters in each detection region along a characteristic direction, determines a cross-sectional fluctuation representation parameter based on the cross-sectional diameter parameters, screens a defect suspected region, obtains a surface image of the defect suspected region and recognizes a line contour, screens a risk contour according to a distribution coefficient of the line contour, marks a first risk sub-contour or determines a characteristic risk contour based on a line representation value of the risk contour, obtains a surface image of an adjacent detection region of the defect suspected region based on a characteristic contour point and recognizes a risk contour, determines a second risk sub-contour and screens a characteristic risk contour according to the characteristic contour point, constructs a defect trend vector according to the risk contour, determines a contour trend representation coefficient, adjusts the illumination parameter for defect detection of the characteristic risk contour based on the contour trend representation coefficient, and performs defect detection of the characteristic risk contour again based on the adjusted illumination parameter, thereby realizing rapid identification of the region with a risk of shadow obstruction, adaptively adjusting the detection mode according to the actual defect characteristics of the shaft part, and improving the precision and reliability of surface defect detection.

[0045] Especially, the present application obtains a plurality of cross-sectional diameter parameters in each detection region along a characteristic direction, and determines a cross-sectional fluctuation representation parameter based on the cross-sectional diameter parameters to screen a defect suspected region. It can be understood that by calculating the adjacent difference value after sorting the cross-sectional diameter parameters of each preset point in the detection region along the characteristic direction, the change of the cross-sectional diameter of the axial part can be captured, thereby representing the degree of mutation of the cross-sectional diameter of the axial part on the illumination path. When the illumination first passes through a region with a larger diameter and then transitions to a region with a smaller diameter, the diameter mutation will cause the reflection angle of the light to change suddenly, forming a local shadow band at the diameter reduction. The gray feature of this shadow band is easy to be confused with surface defects, which can easily lead to misjudgment or missed detection of surface defects. When the cross-sectional fluctuation representation parameter exceeds the cross-sectional fluctuation representation parameter threshold, i.e., the shaft diameter changes significantly, at the same time, the region with cross-sectional diameter change is also the region where the shaft part is prone to surface defects. The present application realizes efficient screening of the defect suspected region, provides target guidance for subsequent accurate detection, effectively improves the pertinence and efficiency of detection, and improves the precision and reliability of surface defect detection.

[0046] Especially, the application screens the risk profile according to the distribution coefficient of the profile contour, and it can be understood that the interval distance between each preset point on the profile contour and the adjacent profile contour is calculated, and the interval distance variance is determined as the distribution coefficient of the profile contour, which can quantify the uniformity and regularity of the profile, accurately identify the abnormal profile through the discrete degree of the profile contour interval distance, effectively distinguish the normal processing profile from the abnormal defect profile, and provide a key screening basis for further analysis of the defect profile, the application screens the risk profile according to the distribution coefficient of the profile contour, and further, the screening of the risk profile is realized, and the precision and reliability of the surface defect detection are improved.

[0047] Especially, the application marks the first risk sub-profile or screens the characteristic risk profile based on the profile characteristic value of the risk profile, and it can be understood that the profile characteristic value is the length characteristic of the risk profile, and by quantifying the length characteristic of the risk profile, it is determined whether the trend of the risk profile can more reliably reflect the real distribution law of the defect, and whether it can be used as a characteristic risk profile for subsequent further analysis, when the profile characteristic value exceeds the threshold value, the profile length of the risk profile is relatively long, and has a certain data representativeness, when the profile characteristic value does not reach the threshold value but the end point is located at the image edge, although the current profile length is insufficient, by marking the first risk sub-profile and combining the adjacent area analysis, the complete morphology of the cross-area defect can be effectively captured, the feature omission caused by the limitation of the single image field of view is avoided, and the information fragmentation problem caused by image segmentation is solved, the application marks the first risk sub-profile or screens the characteristic risk profile based on the profile characteristic value of the risk profile, and further, the screening of the characteristic risk profile and the marking of the first risk sub-profile are realized, and the precision and reliability of the surface defect detection are improved.

[0048] Especially, the application determines the second risk sub-profile according to the characteristic profile point and screens the characteristic risk profile, and it can be understood that by constructing the risk profile correlation mechanism across the images, the defect information fragmentation problem caused by image segmentation is solved, the complete capture and accurate identification of the complex risk profile are realized, the characteristic profile point is used as an anchor point to connect different detection areas, the isolated risk profile segment in the single image is integrated into a complete risk profile morphology, and the total length threshold is determined to ensure that the merged profile has statistical significance, and the misjudgment caused by local noise or short segment interference is avoided, and further, the screening of the characteristic risk profile is realized, and the precision and reliability of the surface defect detection are improved.

[0049] Especially, the application constructs a defect tendency vector according to a risk profile, determines a profile tendency representation coefficient, adjusts an illumination parameter for defect detection of a feature risk profile based on the profile tendency representation coefficient, and can be understood that the defect tendency vector of the feature risk profile is decomposed into an axial and radial direction of a sub-vector, and the ratio thereof is used as a tendency representation coefficient, so that the extension direction of the defect can be determined, when the profile tendency representation coefficient exceeds a threshold range, the illumination is arranged along the radial direction and the angle is adjusted in positive correlation with the coefficient, the shadow contrast of the axial defect is strengthened, when the profile tendency representation coefficient does not reach the threshold, the illumination is arranged along the axial direction and the angle is dynamically adapted, the shadow contrast of the radial defect is highlighted, the maximization highlighting of the defect feature is realized, and then, the detection mode is adaptively adjusted according to the actual defect feature of the shaft part, and the precision and reliability of the surface defect detection are improved. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 A step diagram of a surface detection method for shaft part machining of an embodiment of the application;

[0051] Figure 2 A logic flow diagram for screening a defect suspected area of an embodiment of the application;

[0052] Figure 3 A logic flow diagram for marking a first risk sub-profile or determining a feature risk profile of an embodiment of the application;

[0053] Figure 4 A logic flow diagram for determining a feature profile point of an embodiment of the application. DETAILED DESCRIPTION

[0054] In order to make the objects and advantages of the present application clearer, the present application will be further described below with reference to the 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.

[0055] The preferred embodiments of the present application will be described below with reference to the accompanying drawings. Those skilled in the art should understand 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.

[0056] It should be noted that in the description of the present application, the terms of direction or position relationship such as "upper", "lower", "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 of the present application.

[0057] Moreover, it needs to be explained that, in the description of the present application, unless explicitly defined and limited, the terms "mounting", "connecting" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integral connection, it can be mechanical connection, or electrical connection, it can be direct connection, or indirect connection through intermediate medium, it can be internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0058] Please refer to Figure 1 As shown in the step diagram of the surface detection method for shaft parts processing according to the embodiment of the present application, the surface detection method for shaft parts processing according to the present application comprises:

[0059] In step S100, the surface image of the shaft part to be detected is obtained at a preset light angle and contour recognition is performed, the shaft part is divided into a plurality of detection regions, a plurality of cross-sectional diameter parameters are obtained in each detection region along the characteristic direction, and the cross-sectional fluctuation characteristic parameter is determined based on the cross-sectional diameter parameter to screen the defect suspected area.

[0060] The characteristic direction is determined according to the preset light angle.

[0061] Specifically, the preset light angle can be determined by those skilled in the art according to the ratio of the average value of the cross-sectional diameter of the shaft part to be detected to the length, the preset light angle is negatively correlated with the ratio, the value range of the preset light angle can be [20, 70], the interval unit is °, to avoid that the light angle is too large or too small to affect the detection accuracy, for example, when the average value of the cross-sectional diameter of the shaft part is 80mm, the length is 120mm, and the ratio is 0.67, the preset light angle can be 30°.

[0062] Specifically, the detection region can be uniformly divided into a plurality of detection regions by those skilled in the art according to the axial length and the radial circumference of the shaft part to be detected, the interval distance in the axial direction and the interval angle in the radial direction of each detection region can be set according to the accuracy requirement of surface defect detection, the higher the accuracy requirement, the smaller the interval distance and the interval angle, the value range of the interval distance can be [10, 15], the interval unit is mm, the value range of the interval angle can be [30, 60], the interval unit is °, preferably, the interval distance can be 12mm, and the interval angle can be 45°.

[0063] Specifically, taking the light source as the starting point of the vector and the light beam irradiation direction as the vector direction, the light source irradiation vector is constructed, and the axial direction of the shaft part is orthogonally decomposed along the light source irradiation vector, and the direction of the axial direction vector is the characteristic direction.

[0064] In step S200, a surface image of the defect suspected area is acquired and a grain contour is identified, a risk contour is screened according to a distribution coefficient of the grain contour, and a first risk sub-contour or a feature risk contour is marked or determined based on a grain characterization value of the risk contour.

[0065] Specifically, the surface image of the defect suspected area can be acquired by an industrial camera, and the grain contour can be identified by an edge algorithm, which will not be described again.

[0066] In step S300, a surface image of an adjacent detection area of the defect suspected area is acquired based on the feature contour point, and a risk contour is identified, the surface image containing the feature contour point, a second risk sub-contour is determined according to the feature contour point, and a feature risk contour is screened.

[0067] The feature contour point is determined according to the first risk sub-contour.

[0068] Specifically, the adjacent detection area can be a detection area adjacent to the defect suspected area in the axial direction, or a detection area adjacent to the defect suspected area in the radial direction.

[0069] In step S400, a defect trend vector is constructed according to the risk contour to determine a contour trend characterization coefficient, and a light parameter for defect detection of the feature risk contour is adjusted based on the contour trend characterization coefficient.

[0070] In step S500, the feature risk contour is detected again based on the adjusted light parameter.

[0071] Specifically, determining the cross-section fluctuation characterization parameter includes,

[0072] The cross-section diameter parameters of each preset detection point in the detection area are sorted along the feature direction, the difference between the cross-section diameter parameter of a previous preset detection point and the cross-section diameter parameter of a subsequent preset detection point in adjacent preset detection points is calculated, and the difference is determined as the cross-section fluctuation characterization parameter.

[0073] Specifically, the preset detection points are uniformly distributed along the axial direction of the axial part, the interval distance between adjacent preset detection points is the product of the length of the axial part to be detected and a length factor, the length factor can be set by a person skilled in the art according to the accuracy requirement of surface defect detection, the higher the accuracy requirement, the smaller the length factor, and the length factor can be in the range of [0.02, 0.04], preferably, the length factor can be 0.03.

[0074] Referring to Figure 2 The process of screening the defect suspected area includes,

[0075] If the cross-section fluctuation representation parameter in the detection area meets the defect risk judgment condition, the detection area is screened as a defect suspected area;

[0076] If the cross-section fluctuation representation parameter in the detection area does not meet the defect risk judgment condition, the detection area is not screened.

[0077] The defect risk judgment condition is that at least one cross-section fluctuation representation parameter in the detection area exceeds a preset cross-section fluctuation representation threshold.

[0078] Specifically, the preset cross-section fluctuation representation threshold is the product of a cross-section fluctuation representation reference value and a cross-section fluctuation factor. The cross-section fluctuation representation reference value is the average of the absolute values of the cross-section fluctuation representation parameters of the conventional shaft parts in the historical data. The cross-section fluctuation factor can be set by a person skilled in the art according to the accuracy requirement of surface defect detection. The higher the accuracy requirement, the smaller the cross-section fluctuation factor. The value range of the cross-section fluctuation factor is [1.2, 1.35], and preferably, the cross-section fluctuation factor can be 1.25.

[0079] Specifically, the cross-section diameter parameters in each detection area along the characteristic direction are obtained, and the cross-section fluctuation representation parameter is determined based on the cross-section diameter parameters to screen the defect suspected area. It can be understood that by sorting the cross-section diameter parameters of each preset point in the detection area along the characteristic direction and calculating the adjacent difference, the change of the cross-section diameter of the shaft part can be captured, thereby representing the degree of sudden change of the cross-section diameter of the shaft part on the illumination path. When the light first passes through a region with a larger diameter and then transitions to a region with a smaller diameter, the diameter sudden change will cause the light reflection angle to change suddenly, forming a local shadow band at the diameter reduction. The gray feature of the shadow band is easy to be confused with the surface defect, which can easily lead to misjudgment or missed detection of the surface defect. When the cross-section fluctuation representation parameter exceeds the cross-section fluctuation representation parameter threshold, the shaft diameter changes significantly. At the same time, the region with cross-section diameter change is also the region where the shaft part is prone to have surface defects. The efficient screening of the defect suspected area provides a target guide for subsequent accurate detection, effectively improving the pertinence and efficiency of detection. The embodiment of the present application screens the defect suspected area by the cross-section fluctuation representation parameter, and then realizes the rapid identification of the region with shadow blocking risk, improves the accuracy and reliability of surface defect detection.

[0080] Specifically, it can be understood that the axial diameter mutation on the illumination path will form a local shadow, and the edge of the area with a larger cross-sectional diameter will physically block the area with a smaller cross-sectional diameter on the illumination path, forming a reflection blind area, i.e. a shadow, in the transition area of the axial diameter reduction. The gray value of the shadow area is usually lower than that of the surrounding area, which is easy to be confused with the visual features of defects, and is easy to be misjudged as a surface defect, or to cover the real defects. The diameter difference of adjacent points represents the cross-sectional fluctuation, i.e. quantifies the axial diameter geometric mutation. If the difference is greater than zero, the illumination path first passes through the area with a larger cross-sectional diameter. The greater the difference, the greater the cross-sectional fluctuation parameter, the more obvious the diameter mutation, and the more likely there is shadow interference. The present application embodiment screens the defect suspected area through the cross-sectional fluctuation parameter, and then realizes the rapid identification of the area with the risk of shadow blocking, and improves the accuracy and reliability of surface defect detection.

[0081] Specifically, the process of screening the risk profile includes,

[0082] The interval distance between each preset point on the thread profile and the adjacent thread profile is obtained, and the interval distance variance is determined as the distribution coefficient of the thread profile.

[0083] Specifically, the adjacent thread profile can be the thread profile closest to the current thread profile in the same detection area.

[0084] Specifically, the interval distance between adjacent preset points on the thread profile is the product of the length of the thread profile and the profile length factor. The profile length factor can be set by a person skilled in the art according to the accuracy requirement of surface defect detection. The higher the accuracy requirement, the smaller the profile length factor. The value range of the profile length factor can be [0.1, 0.3], and preferably, the profile length factor can be 0.15.

[0085] Specifically, the interval distance between each preset point on the thread profile and the adjacent thread profile is the vertical interval distance.

[0086] If the distribution coefficient of the thread profile meets the risk profile determination condition, the thread profile is screened as a risk profile.

[0087] If the distribution coefficient of the thread profile does not meet the risk profile determination condition, the thread profile is not screened.

[0088] The risk profile determination condition is that the distribution coefficient exceeds a preset distribution coefficient reference value.

[0089] Specifically, the preset distribution coefficient reference value can be set by a person skilled in the art according to the accuracy requirement of surface defect detection, the higher the accuracy requirement, the smaller the preset distribution coefficient reference value, and the value range of the distribution coefficient reference value can be [0.1, 0.3], preferably, the distribution coefficient reference value can be 0.2.

[0090] Specifically, the embodiment of the present application screens the risk profile according to the distribution coefficient of the groove profile, and it can be understood that the interval distance between each preset point on the groove profile and the adjacent groove profile is calculated, and the interval distance variance is determined as the distribution coefficient of the groove profile, which can quantify the uniformity and regularity of the groove, accurately identify the abnormal profile through the discrete degree of the interval distance of the groove profile, effectively distinguish the normal processing groove from the abnormal defect profile, and provide a key screening basis for further analysis of the defect profile. The embodiment of the present application screens the risk profile according to the distribution coefficient of the groove profile, and further realizes the screening of the risk profile, improves the accuracy and reliability of the surface defect detection.

[0091] Specifically, it can be understood that the surface of the normal shaft part usually has a certain regularity and uniformity, the interval distance between each preset point and the adjacent groove profile is relatively stable, the interval distance variance is small, that is, the distribution coefficient is small, when the surface has a defect profile, the defect profile will destroy this uniformity, the interval distance of the groove profile will change, and the distribution coefficient will be large. The embodiment of the present application screens the risk profile according to the distribution coefficient of the groove profile, and further realizes the screening of the risk profile, improves the accuracy and reliability of the surface defect detection.

[0092] Specifically, please refer to Figure 3 The figure is a logic flow chart for marking the first risk sub-profile or determining the characteristic risk profile according to the embodiment of the present application, and the process of marking the first risk sub-profile or determining the characteristic risk profile includes,

[0093] The interval distance between the end points of the risk profile is determined as the groove characteristic value;

[0094] If the groove characteristic value of the risk profile meets the first characteristic risk profile determination condition, the risk profile is determined as the characteristic risk profile;

[0095] If the groove characteristic value of the risk profile does not meet the first characteristic risk profile determination condition and there is a characteristic groove end point, the risk profile is marked as the first risk sub-profile;

[0096] If the groove characteristic value of the risk profile does not meet the first characteristic risk profile determination condition and there is no characteristic groove end point, the risk profile is not marked;

[0097] The first feature risk profile determination condition is that the texture characteristic value exceeds a preset texture characteristic value threshold.

[0098] Specifically, the texture characteristic value is a maximum value of interval distances between any one end point of the risk profile and the rest end points.

[0099] Specifically, the preset texture characteristic value threshold is a product of a texture characteristic value threshold reference value and a texture characteristic factor, the texture characteristic value threshold reference value is an average value of texture characteristic value thresholds in historical data, and the texture characteristic factor can be set by a person skilled in the art according to an accuracy requirement of surface defect detection. The higher the accuracy requirement is, the larger the texture characteristic factor is. The texture characteristic factor can be in a range of [1.25, 1.4], and preferably, the texture characteristic factor can be 1.35.

[0100] Specifically, the embodiment of the present application marks the first risk sub-profile or screens the feature risk profile based on the texture characteristic value of the risk profile. It can be understood that the texture characteristic value is a length characteristic of the risk profile. By quantifying the length characteristic of the risk profile, it is determined whether the trend of the risk profile can more reliably reflect the real distribution law of defects, and whether it can be used as a feature risk profile for subsequent further analysis. When the texture characteristic value exceeds the threshold value, the profile length of the risk profile is relatively long, and has a certain data representativeness. When the texture characteristic value does not reach the threshold value but the end points are located at the image edge, although the current profile length is insufficient, by marking the first risk sub-profile and combining adjacent area analysis, the complete morphology of the cross-area defect can be effectively captured, feature omission caused by the limitation of the field of view of a single image is avoided, and the information fragmentation problem caused by image segmentation is remedied. The present application marks the first risk sub-profile or screens the feature risk profile based on the texture characteristic value of the risk profile, and further, realizes the screening of the feature risk profile and the marking of the first risk sub-profile, and improves the accuracy and reliability of surface defect detection.

[0101] Specifically, it can be understood that the trend analysis of the risk profile needs enough sample point support, the length of the risk profile can represent the representativeness of the trend data, when the texture representation value exceeds the threshold value, the risk profile contains enough pixel points, the fitting direction vector can more accurately reflect the real extension direction of the defect, and the direction misjudgment probability caused by noise or local interference is reduced, in the surface detection of the shaft type part, the field of view of the single image after image segmentation is limited, and the real defect can cross multiple detection areas, when the end point of the risk profile is located on the image edge, it indicates that the profile can be a local segment of a larger defect, by marking as the first risk sub-profile and triggering the detection of the adjacent area, the related profiles in the adjacent image can be spliced and analyzed by using the feature profile point as an anchor point, the complex trend analysis of all risk profiles is directly performed, a large amount of computing resources is consumed, the feature risk profile with sufficient length is preferentially processed by screening through the texture representation value threshold, the computing resources can be concentrated on the target with the most analysis value, for the first risk sub-profile that does not reach the threshold value but can cross the area, a delayed correlation analysis strategy is adopted, invalid calculation is avoided, and the integrity detection of the complex defect profile is ensured, and then, the screening of the feature risk profile and the marking of the first risk sub-profile are realized, and the precision and reliability of the surface defect detection are improved.

[0102] Specifically, refer to Figure 4 The figure is a logic flow chart for determining the feature profile point in the embodiment of the application, the process of determining the feature profile point comprises,

[0103] If the end point of the first risk sub-profile is on the profile edge of the surface image, the end point is determined as the feature profile point.

[0104] If the end point of the first risk sub-profile is not on the profile edge of the surface image, the end point is not screened.

[0105] Specifically, the process of determining the second risk sub-profile and screening the feature risk profile comprises,

[0106] The risk profile where the feature profile point is located is marked as the second risk sub-profile.

[0107] If the first risk sub-profile and the second risk sub-profile meet the second feature risk profile determination condition, the first risk sub-profile and the second risk sub-profile are jointly determined as the feature risk profile.

[0108] If the first risk sub-profile and the second risk sub-profile do not meet the second feature risk profile determination condition, the first risk sub-profile and the second risk sub-profile are not screened.

[0109] The second feature risk profile determination condition is that the sum of the texture representation value of the first risk sub-profile and the texture representation value of the second risk sub-profile exceeds a preset texture representation value threshold.

[0110] Specifically, according to the feature profile point, the second risk sub-profile is determined, and the feature risk profile is screened. It can be understood that by constructing the risk profile correlation mechanism across the images, the defect information fragmentation problem caused by image segmentation is solved, the complete capture and accurate identification of the complex risk profile are realized, the feature profile point is used as an anchor point to connect different detection regions, the isolated risk profile segment in a single image is integrated into a complete risk profile form, and the length threshold determination is used to ensure that the merged profile has statistical significance, so that misjudgment caused by local noise or short segment interference is avoided. Further, the screening of the feature risk profile is realized, and the accuracy and reliability of the surface defect detection are improved.

[0111] Specifically, it can be understood that in the surface detection of the shaft part, due to the limitation of the camera field of view and the scanning step, a single image can only cover a local area of the part, so that a long defect is easily divided into multiple segments. The feature profile point is the endpoint of the first risk sub-profile at the edge of the image, that is, the extension identifier of the first risk sub-profile in the physical space. Based on the continuity principle of the defect profile in space, that is, the defect profile usually does not suddenly terminate at the edge of the image, but has a continuation in the adjacent region. Taking the point as an anchor point, the risk profile containing the point in the adjacent image is obtained and marked as a second risk sub-profile. The first and second risk sub-profiles are merged to reconstruct the complete form of the profile. Further, the screening of the feature risk profile is realized, and the accuracy and reliability of the surface defect detection are improved.

[0112] Specifically, the process of determining the profile trend representation coefficient includes,

[0113] Taking any endpoint of the feature risk profile as the starting point of the vector, the interval distance between the remaining endpoints on the feature risk profile and the endpoint is calculated, and the endpoint with the maximum interval distance is taken as the terminal point of the vector to construct the defect trend vector.

[0114] The component vector of the defect trend vector in the direction parallel to the axial direction of the shaft part is recorded as a first component vector, and the component vector in the direction perpendicular to the axial direction of the shaft part is recorded as a second component vector.

[0115] The ratio of the module length of the first component vector to the module length of the second component vector is determined as the profile trend representation coefficient.

[0116] Specifically, the defect trend vector can be constructed by calculating the coordinate difference of the feature risk profile pixel point, which is not described again.

[0117] Specifically, the process of adjusting the illumination parameter for defect detection of the feature risk profile includes,

[0118] If the profile trend representation coefficient of the feature risk profile exceeds a preset profile trend representation coefficient threshold range, the illumination direction for defect detection of the feature risk profile is arranged along the radial direction of the shaft part, and the increase amount of the illumination angle is in a positive correlation with the profile trend representation coefficient;

[0119] If the profile trend representation coefficient of the feature risk profile is less than a preset profile trend representation coefficient threshold range, the illumination direction for defect detection of the feature risk profile is arranged along the axial direction of the shaft part, and the increase amount of the illumination angle is in a positive correlation with the profile trend representation coefficient.

[0120] If the profile trend representation coefficient of the feature risk profile is within a preset profile trend representation coefficient threshold range, the illumination direction for defect detection of the feature risk profile is arranged in a conventional illumination direction.

[0121] Specifically, the preset profile trend representation coefficient threshold range can be set by a person skilled in the art according to the accuracy requirement of surface defect detection. The higher the accuracy requirement is, the smaller the preset profile trend representation coefficient threshold range is. The value range of the preset profile trend representation coefficient threshold range can be [0.8, 1.2], and preferably, the profile trend representation coefficient threshold range can be [0.9, 1.1].

[0122] Specifically, when the illumination direction is arranged along the radial direction of the shaft part, the increase amount of the illumination angle is the product of a radial adjustment factor and the profile trend representation coefficient. The radial adjustment factor can be set by a person skilled in the art according to the average value of multiple test data under the same detection condition. The value range of the radial adjustment factor can be [0.8, 0.95] to avoid too much or too little increase amount of the illumination angle, and preferably, the radial adjustment factor can be 0.85.

[0123] Specifically, when the illumination direction is arranged along the axial direction of the shaft part, the increase amount of the illumination angle is the product of an axial adjustment factor and the profile trend representation coefficient. The axial adjustment factor can be set by a person skilled in the art according to the average value of multiple test data under the same detection condition. The value range of the axial adjustment factor can be [1.1, 1.25] to avoid too much or too little increase amount of the illumination angle, and preferably, the axial adjustment factor can be 1.15.

[0124] Specifically, the conventional illumination direction is a commonly set direction of the illumination direction for surface defect detection of the shaft part, and for example, when the average value of the cross-sectional diameter of the shaft part is 80 mm and the length is 120 mm, the illumination direction can be along the radial direction, and the illumination angle can be 30°.

[0125] Specifically, the illumination angle is the included angle between the light source and the tangent plane of the feature risk profile.

[0126] Specifically, the embodiment of the present application constructs a defect trend vector according to the risk profile to determine a profile trend representation coefficient, and adjusts the illumination parameters for defect detection of the feature risk profile based on the profile trend representation coefficient. It can be understood that by decomposing the defect trend vector of the feature risk profile into axial and radial direction vectors and taking the ratio as the trend representation coefficient, the extension direction of the defect can be determined. When the profile trend representation coefficient exceeds the threshold range, the illumination is set along the radial direction and the angle is adjusted in positive correlation with the coefficient to strengthen the shadow contrast of the axial defect. When the profile trend representation coefficient does not reach the threshold, the illumination is set along the axial direction and the angle is dynamically adapted to highlight the shadow contrast of the radial defect, thereby maximizing the highlighting of the defect feature, and further, adaptively adjusting the detection method according to the actual defect feature of the shaft part to improve the accuracy and reliability of surface defect detection.

[0127] Specifically, it can be understood that when the profile trend representation coefficient exceeds the threshold range, the crack tends to be axial, and when the illumination is set along the radial direction, the light is incident at a large angle with the defect trend, and the defect edge forms a significant shadow due to the reflection and shielding of light. The positive correlation between the illumination angle and the profile trend representation coefficient can make the shadow length and contrast increase with the increase of the profile trend representation coefficient, so that the surface defect can be displayed more clearly. When the profile trend representation coefficient is less than the threshold range, the crack tends to be radial, and when the illumination is set along the axial direction, the light is projected along the defect trend, and the concave-convex surface of the defect produces gradient gray scale change due to the difference in optical path. The positive correlation between the illumination angle and the profile trend representation coefficient can make the surface defect appear more clearly, thereby obtaining more accurate surface defect data and improving the accuracy and reliability of surface defect detection.

[0128] So far, the technical solutions of the present application have been described in combination with the preferred embodiments shown in the 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 these changes or replacements will all fall within the protection scope of the present application.

[0129] The above merely illustrates the preferred embodiments of the present application, and is not used to limit the present application; for those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A surface inspection method for machining shaft-type parts, characterized in that, include: The surface image of the shaft part to be inspected is acquired at a preset illumination angle and contour recognition is performed. The shaft part is divided into several inspection areas. Several cross-sectional diameter parameters are acquired in each inspection area along the feature direction. Based on the cross-sectional diameter parameters, cross-sectional fluctuation characterization parameters are determined to screen suspected defect areas. The feature direction is determined according to the preset illumination angle; Acquire surface images of suspected defect areas and identify texture contours. Filter risk contours based on the distribution coefficient of the texture contours. Mark the first risk sub-contour or determine the characteristic risk contour based on the texture characterization value of the risk contour. Based on the feature contour points, the surface image of the adjacent detection area of ​​the suspected defect area is obtained and the risk contour is identified. The surface image contains the feature contour points. The second risk sub-contour is determined and the feature risk contour is filtered according to the feature contour points. The feature contour points are determined based on the first risk sub-contour; A defect tendency vector is constructed based on the risk profile to determine the profile tendency characterization coefficient. The illumination parameters for defect detection of the characteristic risk profile are then adjusted based on the profile tendency characterization coefficient. The process of determining the profile tendency characterization coefficient includes: taking any endpoint of the characteristic risk profile as the vector starting point, calculating the interval distance between the other endpoints on the characteristic risk profile and the endpoint, constructing the defect tendency vector with the endpoint of the maximum interval distance as the vector ending point, and respectively recording the component vector of the defect tendency vector in the axial direction parallel to the shaft part as the first component vector, and the component vector in the axial direction perpendicular to the shaft part as the second component vector, and determining the ratio of the magnitude of the first component vector to the magnitude of the second component vector as the profile tendency characterization coefficient; Defect detection is performed again on the feature risk profile based on the adjusted illumination parameters.

2. The surface inspection method for machining shaft parts according to claim 1, characterized in that, Determining the parameters for characterizing cross-sectional wave fluctuations includes, The cross-sectional diameter parameters of each preset detection point within the detection area are sorted along the characteristic direction. The difference between the cross-sectional diameter parameters of the previous preset detection point and the next preset detection point is calculated, and the difference is determined as the cross-sectional fluctuation characterization parameter.

3. The surface inspection method for machining shaft parts according to claim 2, characterized in that, The process of screening suspected defect areas includes, If the cross-sectional fluctuation characterization parameters within the detection area meet the defect risk determination criteria, then the detection area is screened as a suspected defect area. The defect risk determination condition is that at least one cross-sectional fluctuation characterization parameter in the detection area exceeds the preset cross-sectional fluctuation characterization threshold.

4. The surface inspection method for machining shaft parts according to claim 3, characterized in that, The process of screening risk profiles includes, Obtain the interval distance between each preset point on the texture contour and the adjacent texture contour, and determine the variance of the interval distance as the distribution coefficient of the texture contour; If the distribution coefficient of the texture profile meets the risk profile determination criteria, then the texture profile is selected as a risk profile. The risk profile determination condition is that the distribution coefficient exceeds the preset distribution coefficient reference value.

5. The surface inspection method for machining shaft parts according to claim 4, characterized in that, The process of marking the first risk sub-profile or determining the characteristic risk profile includes, The spacing between the endpoints of the risk profile is determined as the texture characterization value; If the texture representation value of the risk profile meets the first characteristic risk profile determination condition, then the risk profile is determined as a characteristic risk profile. If the texture representation value of the risk contour does not meet the first characteristic risk contour determination condition and there are characteristic texture endpoints, then the risk contour is marked as the first risk sub-contour. The first characteristic risk contour determination condition is that the texture characterization value exceeds the preset texture characterization value threshold.

6. The surface inspection method for machining shaft parts according to claim 5, characterized in that, The process of determining the feature contour points includes, If the endpoint of the first risk sub-contour is on the contour edge of the surface image, then the endpoint is determined as a feature contour point.

7. The surface inspection method for machining shaft parts according to claim 6, characterized in that, The process of determining the second risk sub-profile and screening characteristic risk profiles includes, The risk contour where the feature contour point is located is marked as the second risk sub-contour; If the first risk sub-profile and the second risk sub-profile meet the second characteristic risk profile determination conditions, then the first risk sub-profile and the second risk sub-profile are jointly determined as characteristic risk profiles. The second characteristic risk profile determination condition is that the sum of the texture representation value of the first risk sub-profile and the texture representation value of the second risk sub-profile exceeds a preset texture representation value threshold.

8. The surface inspection method for machining shaft parts according to claim 7, characterized in that, The process of adjusting the illumination parameters for defect detection of the feature risk contour includes, If the contour tendency characterization coefficient of the feature risk contour exceeds the preset contour tendency characterization coefficient threshold range, then the illumination direction for defect detection of the feature risk contour is set along the radial direction of the shaft part, and the increase in the illumination angle is positively correlated with the contour tendency characterization coefficient. If the contour tendency characterization coefficient of the feature risk contour is less than the preset contour tendency characterization coefficient threshold range, then the illumination direction for defect detection of the feature risk contour is set along the axial direction of the shaft part, and the increase in the illumination angle is positively correlated with the contour tendency characterization coefficient.

9. The surface inspection method for machining shaft parts according to claim 8, characterized in that, The illumination angle is the angle between the light source and the tangent of the plane containing the characteristic risk profile.

Images