A quality detection method and system based on machine vision

Abstract

The application relates to a kind of quality detection method and system based on machine vision, relate to the field of cable quality detection, it includes: collection image set, and according to image set generates and sends light irradiation instruction;Thus determine projection data, and determine point image in combination with projection data and sampling point position;Determine variation curve based on point image;According to variation curve and light irradiation instruction determine calculation curve;Determine valley value moment and determine shielding period according to calculation curve;According to shielding period and theoretical period determine time deviation, and determine cumulative duration according to time deviation;Determine variation frequency domain according to variation curve, and the maximum frequency of variation frequency domain is swing frequency;According to swing frequency and working frequency determine frequency deviation, and determine quality deviation in combination with frequency deviation and cumulative duration;When quality deviation is greater than deviation threshold value, generate coating quality warning.The application has the effect of accurately detecting coating quality.

Description

Technical Field

[0001] This invention relates to the field of cable quality inspection, and in particular to a quality inspection method and system based on machine vision. Background Technology

[0002] Quality inspection methods refer to the automated inspection of the material coating process in cable production, which relies on machine vision technology, to accurately identify various quality defects in the coating.

[0003] Existing technologies typically involve using industrial cameras to capture static images of the cable after it has been wrapped, and then using image processing techniques to segment and extract the contours of the static images. This allows for the determination of whether quality defects exist through template matching and morphological analysis.

[0004] Static image information can only reflect the cable status after the coating process is completed, which causes the analysis results of quality defects to lag behind the time when the quality defects occur, making it impossible to detect the coating quality of the cable in real time. Summary of the Invention

[0005] To accurately detect the quality of packaging, this invention provides a quality inspection method and system based on machine vision.

[0006] In a first aspect, the present invention provides a quality inspection method based on machine vision, which adopts the following technical solution: A machine vision-based quality inspection method includes: Step 100: Acquire a set of images, and generate and send a preset light illumination command based on the set of images; Step 101: In response to the light illumination command, determine the projection data of the projection receiving surface, and combine the projection data with the preset sampling points to determine the point image; Step 102: Determine the curve of light intensity change over time at the point based on the image of the point; Step 103: Determine the calculation curve according to the change curve and the light illumination command; Step 104: Determine the valley time based on the calculation curve, and determine the shading period based on the valley time; Step 105: Determine the time deviation based on the occlusion period and the preset theoretical period, and determine the cumulative duration based on the time deviation; Step 106: Determine the frequency domain of the change based on the change curve, and take the maximum frequency of the frequency domain of the change as the oscillation frequency; Step 107: Determine the frequency deviation based on the oscillation frequency and the preset working frequency, and determine the mass deviation by combining the frequency deviation and the cumulative duration; Step 108: When the quality deviation is greater than a preset deviation threshold, a coating quality warning is generated in response to the quality deviation.

[0007] By adopting the above technical solution, the projection light intensity change curve is determined by the image during the cable material coating process, thereby determining the shading period and oscillation frequency, and then calculating the quality deviation, realizing real-time detection of coating material quality defects and improving the accuracy of quality inspection.

[0008] Optionally, the method for correcting the coating quality warning further includes: Step 109: When the quality deviation is not greater than a preset deviation threshold, determine the shadow region based on the image set; Step 110: Determine the shadow outline based on the shadow area; Step 111: Determine the shadow width based on the shadow outline, and determine the width variation based on the shadow width; Step 112: When the width change is greater than a preset change threshold, determine the torsion coefficient according to the width change; Step 113: Determine the edge straightness and edge parallelism of the contour based on the shadow contour, and determine the edge deviation by combining the edge straightness and edge parallelism; Step 114: When the edge deviation is greater than the preset edge threshold, determine the offset coefficient by combining the edge straightness and edge parallelism; Step 115: When the edge deviation is greater than the preset edge threshold and the width change is greater than the preset change threshold, the deformation coefficient is determined according to the torsion coefficient and the offset coefficient, and a covering quality warning is generated based on the deformation coefficient.

[0009] By adopting the above technical solution, the covering material may twist during the cable material covering process. At this time, the shadow width and edge deviation are determined by the shadow outline of the material, thereby determining the twist coefficient and offset coefficient of the covering material, and then updating the covering quality warning, thus improving the accuracy of quality inspection.

[0010] Optionally, it also includes a method for determining the tension coefficient, the method for determining the tension coefficient including: Step 200: When the width change is not greater than a preset change threshold, determine the irradiation direction based on the light irradiation command, and determine the angle between the directions based on the irradiation direction and the covering direction; Step 201: Determine the number of partitions based on the included directional angle; Step 202: Determine the shadow sub-regions by combining the shadow area and the number of partitions, and determine the sub-variation curves based on the shadow sub-regions; Step 203: Determine the sub-mass deviation based on the sub-variation curve; Step 204: Arrange the sub-mass deviations according to the positional order of the shaded sub-regions to form a deviation distribution sequence; Step 205: Determine the deviation gradient based on the deviation distribution sequence; Step 206: When the deviation gradient is greater than a preset gradient threshold, determine the wrapping tension coefficient based on the deviation gradient; Step 207: Generate a coating quality warning based on the tension coefficient.

[0011] By adopting the above technical solution, during the process of cable material covering, the covering material may be subjected to uneven tension. At this time, the shaded area is divided into multiple shaded sub-regions, and the sub-quality deviation is determined for each sub-region, thereby determining the deviation gradient, and then determining the tension coefficient to update the deformation coefficient, which improves the accuracy of quality inspection.

[0012] Optionally, the method for determining the sub-mass deviation further includes: Step 208: When the deviation gradient is greater than a preset gradient threshold, determine the candidate position based on the width change; Step 209: Based on the candidate position, retrieve candidate parameters from edge straightness and edge parallelism, and determine the calculation position according to the candidate parameters; Step 210: Determine the frequency domain segment based on the changed frequency domain and the preset time window; Step 211: Determine the instantaneous frequency based on the frequency domain segment, and determine the sequence of instantaneous frequency changes over time according to the instantaneous frequency; Step 212: Determine the trend slope based on the change sequence; Step 213: Retrieve the calculated slope from the trend slope according to the calculated position; Step 214: Determine the flipping feature based on the calculated slope, and determine the flipping duration according to the flipping feature; Step 215: When the flipping time is less than a preset time threshold, determine the joint time based on the flipping time; Step 216: Determine the joint parameters based on the joint time, and update the sub-mass deviation according to the joint parameters.

[0013] By adopting the above technical solution, the coating material may have joint parts. In this case, the calculation position is determined by the width change, edge straightness and edge parallelism, and the calculation slope is determined by the change frequency domain. Thus, the flipping characteristics and flipping time of the coating material are calculated, and the joint time is determined. Finally, the parameters corresponding to the joint time are removed from the calculation, which improves the accuracy of joint detection.

[0014] Optionally, it also includes a method for generating a warning about abnormal light transmission, the method for generating the warning about abnormal light transmission includes: Step 217: When the tension coefficient is greater than the preset tension threshold, determine the sampling density based on the tension coefficient; Step 218: Determine the internal points according to the sampling density and shadow area, and generate an internal light intensity curve based on the internal points; Step 219: Determine the light transmission deviation based on the internal light intensity curve and the preset reference light transmission curve; Step 220: Determine the light transmission deviation sequence based on the light transmission deviation, and determine the normalized weight in combination with the tension coefficient; Step 221: Determine the comprehensive sequence by combining the normalized weights, the light transmission deviation sequence, and the deviation distribution sequence; Step 222: Determine the comprehensive variance based on the comprehensive sequence, and generate a warning of abnormal light transmission in response to the comprehensive variance.

[0015] By adopting the above technical solution, when the tension on the coating material is too large, it may cause material damage. At this time, the theoretical and actual light transmission deviation is determined based on the light intensity inside the shadow area, and the comprehensive sequence is determined in combination with the quality deviation. Then, the comprehensive variance is determined to generate a light transmission abnormality warning, which improves the accuracy of quality detection.

[0016] Optionally, the method for determining the shaded area further includes: Step 300: Determine the light intensity from the light illumination command, and determine the shadow reproduction degree of each covering material according to the directional angle and the light intensity; Step 301: Determine the final reproduction degree of each wrapping by combining the shadow reproduction degree with the preset roughness of each material, and determine the reference material according to the final reproduction degree; Step 302: Retrieve the reference outline and reference width corresponding to the reference material from the shadow outline and shadow width; Step 303: Determine the spatial pose of the reference material by combining the reference profile and reference width; Step 304: Determine the calculation model according to the spatial pose and the preset mechanism model; Step 305: Determine the simulation area of ​​other coating materials according to the calculation model, and update the shadow area based on the simulation area.

[0017] By adopting the above technical solution, the shadow of the coating material may be distorted. In this case, the final reproduction degree is determined according to the light irradiation command and the coating material. The coating material with the highest final reproduction degree is selected as the reference material. Then, a calculation model is established and the shadow areas of other coating materials are updated, which improves the accuracy of quality inspection.

[0018] Optionally, the method for determining the sampling point location further includes: Step 306: Determine the theoretical light intensity based on the sampling points and calculation model, and determine the actual light intensity based on the shadow area; Step 307: Determine the light intensity ratio by combining the theoretical light intensity and the actual light intensity, and determine the reliability of the location based on the light intensity ratio; Step 308: When the confidence level of the point is greater than the preset untrusted threshold, retrieve the point to be adjusted from the sampled points based on the confidence level of the point. Step 309: Determine the spatial coordinates of the point to be adjusted in the covering material by combining the point to be adjusted and the calculation model, and determine the offset vector by combining the spatial coordinates and the calculation model; Step 310: Update the sampling point position according to the offset vector.

[0019] By adopting the above technical solution, the sampling points of the shadow may be offset. At this time, the theoretical light intensity of each sampling point is determined according to the calculation model, and combined with the actual light intensity to calculate the confidence level of the point. Then, the spatial coordinates of the points with low confidence level on the covering material are calculated. Finally, the sampling points are adjusted to improve the accuracy of the sampling points.

[0020] Optionally, the method for determining the sampling point location further includes: Step 311: When the confidence level of the point is not greater than the preset unconfidential threshold, determine the spatial surface of the coating material based on the calculation model; Step 312: Determine the set of intersection points between the light rays and the covering material based on the spatial curved surface and the irradiation direction; Step 313: Determine the light density distribution on the projection receiving surface by combining the set of intersection points and the calculation model; Step 314: Determine the high-density region by combining the light density distribution and the preset density threshold; Step 315: Determine the reduced adjustment spacing between each sampling point according to the high-density area; Step 316: Update the sampling points according to the adjusted spacing.

[0021] By adopting the above technical solution, when the reliability is high, the surface of the coating material is solved according to the calculation model, and the set of light intersection points and projection surface density distribution of the coating material are obtained. Based on this, high-density areas are delineated and the sampling interval is reduced, thereby realizing point densification optimization and improving the accuracy of quality detection.

[0022] Optionally, the method for determining the sampling point location further includes: Step 317: Determine the region boundary based on the high-density region, and determine the boundary points according to the region boundary; Step 318: Determine the curvature and normal direction of the boundary points according to the boundary points, and determine the vector length based on the curvature of the boundary points; Step 319: Determine the distribution vector by combining the normal direction and the vector length; Step 320: Determine the number of points based on the high-density area and the adjusted spacing, and determine the total length of the vector based on the vector length; Step 321: Determine the distribution length by combining the number of points and the total length of the vector, and update the sampling points by combining the distribution length and the distribution vector.

[0023] By adopting the above technical solution, the distribution vector is determined based on the curvature and normal direction of the boundary points of the high-density area, and the distribution length is calculated based on the adjusted spacing and number of points. This allows the sampling points to be updated adaptively along the boundary normal direction, enabling the sampling points to more accurately fit the edge projection features of the covering material, thereby improving the accuracy of quality inspection.

[0024] Secondly, the present invention provides a quality inspection system based on machine vision, which adopts the following technical solution: The acquisition module is used to acquire a set of images; A memory for storing a program of any one of the machine vision-based quality inspection methods; The processor is the unit of memory that allows programs to be loaded and executed by the processor.

[0025] By adopting the above technical solution, the projection light intensity change curve is determined by the image during the cable material coating process, thereby determining the shading period and oscillation frequency, and then calculating the quality deviation, realizing real-time detection of coating material quality defects and improving the accuracy of quality inspection.

[0026] In summary, the present invention has at least one of the following beneficial technical effects: 1. By determining the projection light intensity change curve through images during the cable material coating process, the shading period and oscillation frequency can be determined, and the quality deviation can be calculated. This enables real-time detection of quality defects in the coating material and improves the accuracy of quality inspection. 2. During the cable material wrapping process, the wrapping material may twist. In this case, the shadow width and edge deviation are determined by the shadow outline of the material, thereby determining the twist coefficient and offset coefficient of the wrapping material, and updating the wrapping quality warning, thus improving the accuracy of quality inspection. 3. During the cable material wrapping process, the wrapping material may be subjected to uneven tension. In this case, the shaded area is divided into multiple shaded sub-regions, and the sub-quality deviation is determined for each sub-region. This determines the deviation gradient, and then the tension coefficient is determined to update the deformation coefficient, thereby improving the accuracy of quality inspection. Attached Figure Description

[0027] Figure 1 This is a flowchart of a quality inspection method based on machine vision. Figure 2 This is a schematic diagram of the light source illumination device; Figure 3 This is a flowchart illustrating the method for determining the tension coefficient; Figure 4 This is a flowchart illustrating the method for determining the shaded area. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0029] This invention discloses a quality inspection method and system based on machine vision.

[0030] Reference Figure 1 and 2 A machine vision-based quality inspection method includes: Step 100: Acquire a set of images, and generate and send a preset light illumination command based on the set of images.

[0031] An image set refers to a collection of image data captured by an industrial camera during the cable wrapping process, including projection data at different time points and images of the wrapping material surface. The image set is acquired by the industrial camera.

[0032] The light illumination command refers to the command that controls the light source to illuminate the cable being wrapped according to preset parameters, including parameters such as light intensity and illumination direction. A timestamp is generated when the light illumination command is sent. In this embodiment, the installation method of the light source illumination device is as follows: the light source and the projection receiving surface are respectively arranged on both sides of the cable being wrapped, forming a backlight projection detection light path. The light source is controlled by the light illumination command, and the light intensity and illumination direction can be adjusted. The projection receiving surface is a planar carrier that receives the shadow of the wrapping material. The wrapping device drives multiple strip-shaped or linear wrapping materials to rotate around the cable, while the cable continues to move along its own wrapping direction. The light source is a point light source, and the initial light illumination command is a light intensity of 1000 Lux, with the illumination direction perpendicular to the wrapping direction.

[0033] Step 101: In response to the light illumination command, determine the projection data of the projection receiving surface, and combine the projection data with the preset sampling points to determine the point image.

[0034] The projection receiving surface refers to the planar carrier used to receive the projection of the covering material. In this embodiment, the projection receiving surface is a white acrylic diffuse reflection projection screen.

[0035] Projection data refers to the image data formed by the covering material on the projection receiving surface. Projection data can be retrieved from the image set. Covering material refers to the protective material used to wrap and cover cables during the cable production process. Covering material is a strip or wire material, and the types of covering materials are included.

[0036] Sampling points refer to fixed observation points on the projection receiving surface used to collect light intensity data. When the center line of the width of the covering material coincides with the line connecting the light source to the center of the receiving surface, the position coordinates of each predefined physical point on the surface of the covering material are converted into the corresponding sampling point coordinates on the projection receiving surface by utilizing the similar triangle projection geometry of the linear propagation of the light source. Each physical point and sampling point is then matched one-to-one. In this embodiment, 20 physical points are predefined at equal intervals on the center line of the width of the covering material, which correspond to 20 sampling points on the projection receiving surface.

[0037] A point image refers to a local image obtained from the projection data according to the sampling point, which is used to analyze the light intensity change at that location. A circular area with the sampling point as the center and the calculated radius as the radius can be extracted from the projection data as the point image. In this embodiment, the calculated radius is 0.1mm.

[0038] Step 102: Determine the curve of light intensity change of the point over time based on the point image.

[0039] A change curve is a continuous curve that reflects the change of light intensity value at a sampling point over time. The average contrast value in the image of the sampling point can be extracted by image processing technology, and the light intensity value corresponding to the contrast value can be found from the light intensity value correspondence table. The light intensity value correspondence table is a data table that records different contrast values ​​and their corresponding light intensity values. Then, the light intensity values ​​at each time are sorted according to time to fit and form a change curve.

[0040] Through calibration experiments, with fixed optical path and camera parameters, a known gradient light intensity is output from a light source, shadow images are captured, and the root mean square contrast of the sampling area is calculated. The contrast and light intensity data are then mapped through interpolation or fitting to generate a light intensity value correspondence table for online detection and query. In this embodiment, the correspondence tables for quality deviation, torsion coefficient, offset coefficient, number of partitions, tension coefficient, sampling density, normalized weight, shadow reproduction degree, roughness, correction coefficient, inherent parameter, adjustment spacing, and vector length can all be established according to the calibration experiment method.

[0041] Step 103: Determine the calculation curve according to the change curve and the light illumination command.

[0042] The calculation curve refers to the segment of the change curve used in the calculation process, that is, the part of the change curve after the light irradiation command is emitted. The calculation curve can be extracted from the change curve according to the timestamp of the light irradiation command.

[0043] Step 104: Determine the valley time based on the calculation curve, and determine the occlusion period based on the valley time.

[0044] The valley moment refers to the time point on the calculation curve where the light intensity value reaches a local minimum. The valley moment can be determined from the calculation curve using an extreme value search algorithm.

[0045] The occlusion period refers to the periodic time interval between the covering material blocking the light source. The time difference between two adjacent trough moments is the occlusion period.

[0046] Step 105: Determine the time deviation based on the occlusion period and the preset theoretical period, and determine the cumulative duration based on the time deviation.

[0047] The theoretical period refers to the standard period during which the coating material blocks the light source under ideal conditions. The equipment wraps the coating material at its operating speed. At this time, the coating material blocks the light source once for each rotation. The rotation period of the coating material is the theoretical period. In this embodiment, the theoretical period is 0.04s.

[0048] Time deviation refers to the difference between the actual shading period and the theoretical period. The absolute value of the difference between the shading period and the theoretical period can be used as the time deviation to reflect the deviation generated by the coating material in a single period.

[0049] Cumulative duration refers to the sum of all time deviations, used to reflect the cumulative periodic deviations of the coating material during long-term operation.

[0050] Step 106: Determine the frequency domain of the change based on the change curve, and take the maximum frequency of the frequency domain of the change as the oscillation frequency.

[0051] The variable frequency domain refers to the frequency domain of the shaded area, and the variable curve can be transformed into the variable frequency domain through Fourier transform.

[0052] The oscillation frequency refers to the frequency at which the coating material oscillates during the coating process. The frequency corresponding to the frequency component with the highest energy in the changing frequency domain can be taken as the oscillation frequency.

[0053] Step 107: Determine the frequency deviation based on the oscillation frequency and the preset working frequency, and determine the mass deviation by combining the frequency deviation and the cumulative duration.

[0054] The operating frequency refers to the operating frequency of the coating equipment under standard working conditions. When the equipment is wrapping the coating material, the motor will vibrate. This vibration will be transmitted to the coating material through the wrapping head. The corresponding frequency is the operating frequency. In this embodiment, the operating frequency is 25Hz.

[0055] Frequency deviation refers to the difference between the actual working frequency and the theoretical working frequency of the coating material. Under actual working conditions, the oscillation frequency of the coating material may be greater than the theoretical working frequency due to interference from external factors. The difference between the oscillation frequency and the working frequency can be used as the frequency deviation to reflect the severity of the disturbance to the coating material during processing.

[0056] Quality deviation refers to the stability deviation value of the coating material. The larger the frequency deviation, the more severe the disturbance to the coating material. The longer the cumulative duration, the longer the disturbance lasts, and the larger the quality deviation. The quality deviation corresponding to the frequency deviation and cumulative duration can be found in the quality deviation correspondence table. The quality deviation correspondence table is a data table that records different frequency deviations and cumulative durations and their corresponding quality deviations. In this embodiment, under actual working conditions, the interference of the external environment will cause the stability of the cable coating process to be lower than the theoretical state, resulting in a quality deviation greater than zero.

[0057] Step 108: When the quality deviation is greater than a preset deviation threshold, a coating quality warning is generated in response to the quality deviation.

[0058] The deviation threshold is a critical value used to determine whether the coating quality is qualified. When the quality deviation is greater than the deviation threshold, the coating quality is unqualified. In this embodiment, the deviation threshold is 0.2.

[0059] A wrapping quality warning is a warning that alerts workers to defects in the wrapping material and is displayed on the screen.

[0060] By determining the projection light intensity change curve through images during the cable material coating process, the shading period and oscillation frequency can be determined, and the quality deviation can be calculated. This enables real-time detection of quality defects in the coating material and improves the accuracy of quality inspection.

[0061] Correction methods for coating quality warnings also include: Step 109: When the quality deviation is not greater than a preset deviation threshold, determine the shadow area based on the image set.

[0062] The shadow region refers to the set of pixels in the shadow part of the image set, that is, the projection area formed by the covering material on the projection receiving surface. Pixels with a contrast lower than a preset contrast threshold can be segmented from the point image by image recognition technology, thereby forming a continuous and closed area as the shadow region. In this embodiment, the contrast threshold is 30.

[0063] Step 110: Determine the shadow outline based on the shadow area.

[0064] The shadow outline refers to the outer boundary outline of the shadow area. The closed curve formed by connecting the outermost pixels of the shadow area can be used as the shadow outline.

[0065] Step 111: Determine the shadow width based on the shadow outline, and determine the width variation based on the shadow width.

[0066] The shadow width refers to the maximum span of the shadow outline perpendicular to the covering direction. The covering direction refers to the direction in which the covering material travels on the cable. The outline width corresponding to the sampling point can be determined from the shadow outline through a geometric measurement algorithm. In this embodiment, the covering direction is the direction in which the cable travels, and the direction in which the cable travels is from left to right.

[0067] Width variation refers to the change in the width of the shadow at adjacent sampling points. It is used to reflect the uniformity of the width of the covering material. The absolute value of the difference between the shadow width corresponding to the sampling point and the shadow width corresponding to the previous sampling point can be used as the width variation.

[0068] Step 112: When the width change is greater than the preset change threshold, determine the torsion coefficient according to the width change.

[0069] The change threshold is a critical value used to determine whether the width change is abnormal. When the width change is greater than the change threshold, the covering material may be torsional. In this embodiment, the change threshold is 0.1 mm.

[0070] The torsion coefficient is a quantitative value that reflects the degree of torsion of the covering material. The greater the width change, the more severe the deformation of the covering material, and the greater the torsion coefficient. The torsion coefficient corresponding to the width change can be found in the torsion coefficient correspondence table, which is a data table that records different width changes and their corresponding torsion coefficients.

[0071] Step 113: Determine the edge straightness and edge parallelism of the contour based on the shadow contour, and determine the edge deviation by combining the edge straightness and edge parallelism.

[0072] Edge straightness refers to the degree of deviation between the edge of the shadow contour and the ideal straight line. It can be achieved by fitting the edge straight line using the least squares method and calculating the edge distance from each point on the shadow contour to the edge straight line, and then taking the average edge distance as the edge straightness.

[0073] Edge parallelism refers to the degree of parallelism between the two edges of a shadow outline. It can be calculated by taking the angle between the straight lines of the two edges.

[0074] Edge deviation refers to a quantitative value reflecting the regularity of the shadow contour edge. The contour length of the shadow contour along the covering direction is calculated by image recognition technology, and the parallel value is calculated according to the formula parallel value = contour length * tan (edge ​​parallelism). The weighted average of edge straightness and parallel value is used as the edge deviation. In this embodiment, the weight of edge straightness is 0.5 and the weight of parallel value is 0.5.

[0075] Step 114: When the edge deviation is greater than the preset edge threshold, the offset coefficient is determined by combining the edge straightness and edge parallelism.

[0076] The edge threshold is a critical value used to determine whether the edge deviation is abnormal. When the edge deviation is greater than the edge threshold, the edge of the covering material is distorted. In this embodiment, the edge threshold is 0.15mm.

[0077] The offset coefficient is a quantitative value that reflects the degree of offset of the covering material. The larger the edge deviation, the more serious the deformation of the shaded area, and the larger the offset coefficient. The offset coefficient corresponding to the edge deviation can be found in the offset coefficient correspondence table. The offset coefficient correspondence table is a data table that records different edge deviations and their corresponding offset coefficients.

[0078] Step 115: When the edge deviation is greater than the preset edge threshold and the width change is greater than the preset change threshold, the deformation coefficient is determined according to the torsion coefficient and the offset coefficient, and a covering quality warning is generated based on the deformation coefficient.

[0079] The deformation coefficient is a quantitative coefficient that reflects the degree of deformation of the coating material. The product of the torsional coefficient and the offset coefficient can be used as the deformation coefficient, and the product of the original mass deviation and the deformation coefficient can be used as the new mass deviation.

[0080] During the cable material wrapping process, the wrapping material may twist. In this case, the shadow width and edge deviation are determined by the shadow outline of the material, thereby determining the twist coefficient and offset coefficient of the wrapping material, and then updating the wrapping quality warning, thus improving the accuracy of quality inspection.

[0081] Reference Figure 3 The methods for determining the tension coefficient include: Step 200: When the width change is not greater than a preset change threshold, determine the irradiation direction based on the light irradiation command, and determine the directional angle based on the irradiation direction and the covering direction.

[0082] The irradiation direction refers to the direction of light propagation when the light source irradiates the covering material. The irradiation direction can be retrieved from the light irradiation command.

[0083] The directional angle is the angle between the direction of illumination and the direction of coverage, used to measure the magnification of shadows caused by the angle of incidence of light.

[0084] Step 201: Determine the number of partitions based on the included directional angle.

[0085] The number of partitions refers to the width of each sub-region when dividing the shaded area. When the directional angle is close to 90 degrees, it means that the area of ​​the shaded area is closer to the actual cross-sectional area of ​​the cable, and the number of partitions is less. The number of partitions corresponding to the directional angle can be found in the partition number correspondence table, which is a data table that records different directional angles and their corresponding number of partitions.

[0086] Step 202: Determine the shadow sub-regions by combining the shadow area and the number of partitions, and determine the sub-variation curves based on the shadow sub-regions.

[0087] A shadow sub-region refers to a shadow region that is continuously and equidistantly divided into multiple long strip-shaped regions along the short axis of the shadow region according to the number of partitions. These long strip-shaped regions are called shadow sub-regions.

[0088] The sub-variation curve refers to the curve of light intensity change over time within each shadow sub-region, which can be determined by referring to the method for determining the variation curve in steps 101 to 102.

[0089] Step 203: Determine the sub-mass deviation based on the sub-variation curve.

[0090] Sub-quality deviation refers to the quality deviation value corresponding to a single shaded sub-region, which can be determined by referring to the quality deviation determination method in steps 104 to 107.

[0091] Step 204: Arrange the sub-quality deviations according to the positional order of the shaded sub-regions to form a deviation distribution sequence.

[0092] The deviation distribution sequence refers to the sequence formed by arranging the sub-mass deviations of each shaded sub-region along a direction perpendicular to the covering direction, which is used to reflect the spatial distribution characteristics of the mass deviation.

[0093] Step 205: Determine the deviation gradient based on the deviation distribution sequence.

[0094] The deviation gradient refers to the change in value between adjacent sub-mass deviations in the deviation distribution sequence. It is used to reflect the spatial distribution trend of mass deviations. The absolute value of the difference between a sub-mass deviation and the previous sub-mass deviation can be used as the deviation gradient corresponding to that sub-mass deviation.

[0095] Step 206: When the deviation gradient is greater than a preset gradient threshold, determine the tension coefficient of the coating based on the deviation gradient.

[0096] The gradient threshold is a critical value used to determine whether the deviation gradient is abnormal. When the deviation gradient is greater than the gradient threshold, it means that the sub-mass deviation in each region has a significant abrupt change in space, reflecting that the dynamic response of different parts of the material is inconsistent due to the difference in force. Therefore, the coating material may have uneven tension.

[0097] The tension coefficient is a quantitative value that reflects the degree of unevenness of the tension on the coating material. The larger the deviation gradient, the greater the quality difference between the sub-regions and the greater the tension change, the larger the tension coefficient. The tension coefficient corresponding to the deviation gradient can be found in the tension coefficient correspondence table, which is a data table that records different deviation gradients and their corresponding tension coefficients.

[0098] Step 207: Generate a coating quality warning based on the tension coefficient.

[0099] During the cable material sheathing process, the sheathing material may be subjected to uneven tension. In this case, the shaded area is divided into multiple shaded sub-regions, and the sub-quality deviation is determined for each sub-region. This determines the deviation gradient, and then the tension coefficient is determined to update the deformation coefficient, thereby improving the accuracy of quality inspection.

[0100] Methods for determining sub-mass deviation also include: Step 208: When the deviation gradient is greater than the preset gradient threshold, determine the candidate position based on the width change.

[0101] Candidate locations refer to the points where the shadow width changes abnormally. Sampling points where the width change exceeds the change threshold can be used as candidate locations.

[0102] Step 209: Based on the candidate position, retrieve candidate parameters from edge straightness and edge parallelism, and determine the calculation position according to the candidate parameters.

[0103] Candidate parameters refer to the edge straightness and edge parallelism values ​​corresponding to the candidate position. The corresponding data can be retrieved from the edge straightness and edge parallelism values ​​based on the candidate position as candidate parameters.

[0104] The calculation position refers to the position where the candidate parameters in the candidate position show a synchronization anomaly. The edge deviation can be calculated based on the candidate parameters. When the edge deviation is greater than the edge threshold, the candidate position corresponding to the candidate parameter is used as the calculation position.

[0105] Step 210: Determine the frequency domain segment based on the changed frequency domain and the preset time window.

[0106] The time window refers to the preset time period length used to capture the changing frequency domain. In this embodiment, the time window is 0.2s.

[0107] A frequency domain segment refers to local frequency domain data extracted from a variable frequency domain according to a time window, which is used to calculate whether there is a joint in the covering material.

[0108] Step 211: Determine the instantaneous frequency based on the frequency domain segment, and determine the sequence of instantaneous frequency changes over time according to the instantaneous frequency.

[0109] Instantaneous frequency refers to the frequency value corresponding to each moment in a frequency domain segment. The frequency component with the highest energy in the frequency domain segment can be determined as the instantaneous frequency through time-frequency analysis algorithms.

[0110] A change sequence refers to a sequence of data in which the instantaneous frequency changes over time. The instantaneous frequencies at each moment can be arranged in chronological order to form a change sequence.

[0111] Step 212: Determine the trend slope based on the change sequence.

[0112] The trend slope refers to the slope of the frequency change trend in a changing sequence. It can be calculated by comparing the slope of the instantaneous frequency in the changing sequence with the slope of the previous instantaneous frequency as the trend slope of that instantaneous frequency.

[0113] Step 213: Retrieve the calculated slope from the trend slope according to the calculated position.

[0114] The calculated slope refers to the trend slope value corresponding to the calculation position. The calculated slope can be read from the trend slope based on the calculation position.

[0115] Step 214: Determine the flipping feature based on the calculated slope, and determine the flipping duration according to the flipping feature.

[0116] The flip feature refers to the characteristic pattern of the instantaneous frequency that first increases in the positive direction and then decreases in the negative direction when the joint of the covering material passes through the projection receiving surface. When the calculated slope suddenly changes from a positive value to a negative value, the flip feature is determined to exist.

[0117] The flip duration refers to the length of time a flip feature lasts. The difference between the end time and the start time of the flip feature can be used as the flip duration.

[0118] Step 215: When the flipping time is less than a preset time threshold, determine the joint time based on the flipping time.

[0119] The duration threshold is the critical time used to determine whether the flipping feature corresponds to the flipping of the joint. The coating material joint refers to the physical joint formed when two broken coating materials are connected together. The physical length of the coating material joint is small, and the time to pass through the sampling point is short. Therefore, when the flipping time is less than the duration threshold, it is determined to be a joint. In this embodiment, the duration threshold is 0.005s.

[0120] The joint time refers to the specific time range during which the joint of the covering material passes through the sampling point. The time range enclosed by the calculation position where the flipping feature appears can be used as the joint time.

[0121] Step 216: Determine the joint parameters based on the joint time, and update the sub-mass deviation according to the joint parameters.

[0122] Joint parameters refer to the range of data that needs to be removed from the calculation at the joint time. The change curve corresponding to the joint time can be used as the joint parameters, and the joint parameters can be removed from the change curve.

[0123] The coating material may have joints. In this case, the calculation location is determined by the width variation, edge straightness and edge parallelism, and the calculation slope is determined by the frequency domain variation. This allows us to calculate the flipping characteristics and flipping time of the coating material, thereby determining the joint time. Finally, the parameters corresponding to the joint time are removed from the calculation, which improves the accuracy of joint detection.

[0124] Methods for generating warnings about abnormal light transmission include: Step 217: When the tension coefficient is greater than the preset tension threshold, determine the sampling density based on the tension coefficient.

[0125] Tension threshold refers to the critical value used to determine whether the tension is too high. When the tension coefficient is greater than the tension threshold, the covering material may be overstretched, resulting in light transmission.

[0126] Sampling density refers to the density of points when sampling light intensity within a shaded area. The higher the tension coefficient, the higher the sampling density. The sampling density corresponding to the tension coefficient can be found in the sampling density correspondence table, which is a data table that records different tension coefficients and their corresponding sampling densities.

[0127] Step 218: Determine the internal points according to the sampling density and shadow area, and generate an internal light intensity curve based on the internal points.

[0128] Internal points refer to light intensity sampling points that are uniformly distributed according to sampling density within the shaded area. The shaded area can be divided into grids according to sampling density, and the intersection of the grid lines are the internal points. Internal points can be added to the original sampling points to calculate the internal light intensity curve.

[0129] The internal light intensity curve refers to the curve showing how the light intensity at an internal point changes over time. It can be determined by referring to the method for determining the change curve in step 102.

[0130] Step 219: Determine the light transmission deviation based on the internal light intensity curve and the preset reference light transmission curve.

[0131] Transmission deviation refers to the difference between the actual internal light intensity curve and the reference transmission curve. The mean square error between the two can be calculated as the transmission deviation.

[0132] Step 220: Determine the light transmission deviation sequence based on the light transmission deviation, and determine the normalized weight in combination with the tension coefficient.

[0133] The light transmission deviation sequence refers to the sequence formed by arranging the light transmission deviations corresponding to all internal points in order of position, and is used to calculate whether the covering material has abnormal light transmission.

[0134] Normalization weight refers to the weighting coefficient used to normalize the light transmission deviation sequence and the deviation distribution sequence. The larger the tension coefficient, the tighter the covering material, and the larger the normalization weight. The normalization weight corresponding to the tension coefficient can be found in the normalization weight correspondence table, which is a data table that records different tension coefficients and their corresponding normalization weights.

[0135] Step 221: Determine the comprehensive sequence by combining the normalized weights, the light transmission deviation sequence, and the deviation distribution sequence.

[0136] The comprehensive sequence refers to the comprehensive evaluation sequence formed by superimposing the weighted transmission deviation sequence and the deviation distribution sequence at corresponding positions.

[0137] Using the coordinates of each element in the deviation distribution sequence as a reference, linear interpolation is performed on the original light transmission deviation sequence to obtain an aligned light transmission deviation sequence. Then, each value in the light transmission deviation sequence and the deviation distribution sequence is normalized. The normalized deviation distribution value is then multiplied by the normalized weight and the normalized light transmission deviation value as a superposition term. The product of the normalized deviation distribution value and the normalized light transmission deviation value is then calculated as a cooperative cross term. The superposition term and the cooperative cross term are then added together to obtain the comprehensive sequence value at that position. By performing this calculation point by point for all positions, a comprehensive sequence corresponding one-to-one with the position of the shadow sub-region is formed.

[0138] Step 222: Determine the comprehensive variance based on the comprehensive sequence, and generate a warning of abnormal light transmission in response to the comprehensive variance.

[0139] The overall variance refers to the degree of dispersion of each element in the overall sequence from its mean. The variance of the overall sequence can be used as the overall variance to reflect the uniformity of the light transmission performance of the coating material. The larger the overall variance, the more uneven the light transmission performance.

[0140] The warning of abnormal light transmission in the covering material indicates that the covering material may have local damage, thinning, or other abnormal light transmission, and the warning will be displayed on the screen.

[0141] When the coating material is subjected to excessive tension, it may be damaged. In this case, the theoretical and actual light transmission deviation is determined based on the light intensity inside the shadow area, and a comprehensive sequence is determined in combination with the quality deviation. Then, the comprehensive variance is determined to generate a light transmission anomaly warning, which improves the accuracy of quality inspection.

[0142] Reference Figure 4 Methods for determining the shaded area also include: Step 300: Determine the light intensity from the light illumination command, and determine the shadow reproduction degree of each covering material according to the directional angle and the light intensity.

[0143] Light intensity refers to the numerical value of the light emitted by a light source, which can be retrieved from the light illumination command.

[0144] Shadow fidelity refers to the degree to which the shadow area matches the shape of the actual covering material under specific lighting conditions. The closer the directional angle is to 90 degrees and the greater the light intensity, the higher the shadow fidelity. The shadow fidelity corresponding to the directional angle and light intensity can be found in the shadow fidelity correspondence table. The shadow fidelity correspondence table is a data table that records different directional angles and light intensities and their corresponding shadow fidelities.

[0145] Step 301: Combine the shadow reproduction degree with the preset roughness of each material to determine the final reproduction degree of each wrapping, and determine the reference material according to the final reproduction degree.

[0146] Roughness refers to the roughness parameter of the boundary of the coating material, which affects the shadow imaging effect of the coating material. The roughness of the coating material can be found from the preset roughness correspondence table, which is a data table that records different coating materials and their corresponding roughness.

[0147] Final fidelity refers to the overall fidelity index of shadows. The greater the roughness, the more chaotic the boundary of the covering material's blocking effect on light, resulting in a worse shadow imaging effect and a smaller correction coefficient. The correction coefficient corresponding to the roughness can be found in the correction coefficient correspondence table. The product of the shadow fidelity and the correction coefficient is then used as the final fidelity. The correction coefficient correspondence table is a data table that records different roughness levels and their corresponding correction coefficients.

[0148] The reference material refers to the coating material with the highest final fidelity, which serves as a reference standard for subsequent shadow correction.

[0149] Step 302: Retrieve the reference outline and reference width corresponding to the reference material from the shadow outline and shadow width.

[0150] The reference profile refers to the shadow profile corresponding to the reference material, and the reference width refers to the shadow width corresponding to the reference material. It can be retrieved from the shadow profile and shadow width based on the reference material.

[0151] Step 303: Determine the spatial pose of the reference material by combining the reference profile and reference width.

[0152] Spatial pose refers to the position and orientation parameters of a reference material in three-dimensional space, including spatial coordinates and rotation angles. The spatial pose can be determined by geometric analysis algorithms based on the reference contour, reference width, and preset inherent parameters of the reference material. The inherent parameters of the covering material can be retrieved from the preset inherent parameter correspondence table. The inherent parameter correspondence table is a data table that records different covering materials and their corresponding inherent parameters. The inherent parameters include the geometric dimensions and working angles of the reference material.

[0153] Step 304: Determine the calculation model according to the spatial pose and the preset mechanism model.

[0154] The mechanism model refers to the mechanical structure and spatial layout model of the covering equipment, which can be determined according to the installation method of the light source illumination device.

[0155] The computational model refers to the theoretical model used to calculate the shadow correction parameters of other covering materials. The computational model is used to deduce the simulated area of ​​other covering materials and replace the original shadow area. Spatial pose can be used as the initial condition and substituted into the mechanism model to establish a set of projection transformation relationships from the actual covering material to the shadow of the projection receiving surface. Finally, a computational model that can deduce the theoretical shadow outline and width of other covering materials is generated.

[0156] Step 305: Determine the simulation area of ​​other coating materials according to the calculation model, and update the shadow area based on the simulation area.

[0157] The shadows cast by the coating material may be distorted. In this case, the final reproduction degree is determined based on the light illumination command and the coating material. The coating material with the highest final reproduction degree is then selected as the reference material. A calculation model is then established and the shadow areas of other coating materials are updated, which improves the accuracy of quality inspection.

[0158] Methods for determining sampling points also include: Step 306: Determine the theoretical light intensity based on the sampling points and calculation model, and determine the actual light intensity based on the shadow area.

[0159] Theoretical light intensity refers to the standard light intensity value at the sampling point location calculated based on the calculation model.

[0160] Actual light intensity refers to the actual light intensity value at the sampling point in the shaded area. The value of the change curve at the current moment of the sampling point can be read as the actual light intensity.

[0161] Step 307: Determine the light intensity ratio by combining the theoretical light intensity and the actual light intensity, and determine the reliability of the location based on the light intensity ratio.

[0162] The light intensity ratio refers to the ratio of the actual light intensity to the theoretical light intensity. The quotient of the actual light intensity and the theoretical light intensity can be used as the light intensity ratio.

[0163] Location reliability refers to the quantitative value of the reliability of sampling point data. The absolute value of the difference between the light intensity ratio and 1 can be used as the location reliability.

[0164] Step 308: When the confidence level of the point is greater than the preset untrusted threshold, retrieve the point to be adjusted from the sampled points based on the confidence level of the point.

[0165] The unreliability threshold is a critical value used to determine whether the reliability of a sampling point meets the standard. When the reliability of a sampling point is greater than the unreliability threshold, there is a significant deviation between the actual measured value and the theoretical predicted value. At this time, the location of the sampling point may have drifted, and the actual collected data may be distorted.

[0166] Points to be adjusted refer to sampling points whose reliability exceeds the unreliability threshold and need to be adjusted. Points to be adjusted need to undergo position correction.

[0167] Step 309: Determine the spatial coordinates of the point to be adjusted in the covering material by combining the point to be adjusted and the calculation model, and determine the offset vector by combining the spatial coordinates and the calculation model.

[0168] Spatial coordinates refer to the coordinate position of the point to be adjusted in the three-dimensional space of the covering material. When the width center line on the covering material coincides with the line connecting the light source to the center of the receiving surface, the position coordinates of the point to be adjusted can be converted into spatial coordinates on the covering material by using the similar triangle projection geometric relationship of the linear propagation of the light source.

[0169] The offset vector refers to the offset direction and distance between the actual position and the theoretical position of the point to be adjusted. It can be calculated as the distance vector between the spatial coordinates and the position coordinates. The position coordinates are the position coordinates of the physical point corresponding to the point to be adjusted. The distance vector is mapped onto the projection receiving surface as the offset vector, and the point to be adjusted is moved according to the offset vector as the new sampling point.

[0170] Step 310: Update the sampling point position according to the offset vector.

[0171] The sampling points of the shadow may be offset. In this case, the theoretical light intensity of each sampling point is determined according to the calculation model and combined with the actual light intensity to calculate the confidence level of the point. Then, the spatial coordinates of the points with low confidence level on the covering material are calculated. Finally, the sampling points are adjusted to improve the accuracy of the sampling points.

[0172] Methods for determining sampling points also include: Step 311: When the credibility of the point is not greater than the preset untrusted threshold, determine the spatial surface of the covering material based on the calculation model.

[0173] A spatial surface refers to the surface shape of a covering material in three-dimensional space, and a spatial surface can be determined by fitting a computational model.

[0174] Step 312: Determine the set of intersection points between the light rays and the covering material based on the spatial curved surface and the irradiation direction.

[0175] The intersection set refers to the set of all intersection points when light shines on the surface of the coating material. The intersection set can be determined in the computational model by using a ray tracing algorithm based on the spatial surface and the direction of illumination.

[0176] Step 313: Determine the light density distribution on the projection receiving surface by combining the set of intersection points and the calculation model.

[0177] Light density distribution refers to the distribution of light at various locations on the projection receiving surface. The process of light illuminating the three-dimensional curved surface of the covering material is simulated using a ray tracing algorithm, and the light density distribution at various locations on the projection receiving surface is calculated through the set of intersection points.

[0178] Step 314: Determine the high-density region by combining the light density distribution and the preset density threshold.

[0179] The density threshold is a critical value used to define high-density areas. Areas with light density greater than the density threshold may experience more drastic changes in light intensity.

[0180] High-density areas refer to regions in the light density distribution where the light density exceeds a density threshold. Sampling points in high-density areas need to be densified.

[0181] Step 315: Determine the reduced adjustment spacing between each sampling point according to the high-density area.

[0182] The adjustment spacing refers to the reduced spacing between sampling points in a high-density area. The higher the light density, the more drastic the curvature change of the coating material surface. The light intensity here is more sensitive to changes in spatial position, and the signal is more likely to jump due to small distances. Therefore, the reduction distance is larger. The adjustment spacing corresponding to the light density can be found in the adjustment spacing correspondence table. The adjustment spacing correspondence table is a data table that records different light densities and their corresponding adjustment spacings. The original sampling point spacing is then subtracted from the reduction distance to obtain the densified spacing, and sampling points are regenerated in the high-density area. In this embodiment, the original sampling point spacing is 1 mm.

[0183] Step 316: Update the sampling points according to the adjusted spacing.

[0184] When the reliability is high, the surface of the coating material is solved according to the calculation model, and the set of light intersection points and projection surface density distribution of the coating material are obtained. Based on this, high-density areas are delineated and the sampling interval is reduced to achieve point densification optimization and improve the accuracy of quality detection.

[0185] Methods for determining sampling points also include: Step 317: Determine the region boundary based on the high-density region, and determine the boundary points according to the region boundary.

[0186] A region boundary refers to the dividing line between a high-density region and its surrounding regions. It can be defined as a closed curve formed by connecting the outermost pixels of the high-density region.

[0187] Boundary points refer to points located on the boundary of a region, used to capture the fine features of the projection of the edge of the covering material. The boundary of the region can be divided equally according to a preset arc length spacing, and each dividing point is a boundary point. In this embodiment, the arc length spacing is 0.05mm.

[0188] Step 318: Determine the curvature and normal direction of the boundary points, and determine the vector length based on the curvature of the points.

[0189] Point curvature refers to the degree of curvature of the boundary of a region at a boundary point. It can be calculated by numerical differentiation as the second derivative of the boundary profile at the boundary point.

[0190] The normal direction refers to the direction at the boundary point that is perpendicular to the boundary tangent and points into the high-density region. The boundary tangent at the boundary point can be determined by the central difference method combined with short arc line fitting, and the direction that is perpendicular to the boundary tangent and points into the high-density region is taken as the normal direction.

[0191] Vector length refers to the offset distance determined by the curvature of a point, which is used to control the distribution range of newly added sampling points. The greater the curvature of a point, the more curved the boundary is at that point, and the more drastic the spatial change of the projection features. More sampling points need to be arranged to collect information, so the vector length is larger. The vector length corresponding to the curvature of a point can be found in the vector length correspondence table, which is a data table that records different point curvatures and their corresponding vector lengths.

[0192] Step 319: Determine the distribution vector by combining the normal direction and the vector length.

[0193] The distribution vector is a spatial vector used to indicate the direction and distance of sampling point placement. It can be the product of the unit vector of the normal direction and the vector length.

[0194] Step 320: Determine the number of points based on the high-density area and the adjusted spacing, and determine the total length of the vector based on the vector length.

[0195] The number of sampling points refers to the total number of sampling points in a high-density area, used to calculate the distribution length.

[0196] The total vector length refers to the sum of the vector lengths corresponding to all boundary points, which is used to reflect the complexity of the overall curvature of the boundary.

[0197] Step 321: Determine the distribution length by combining the number of points and the total length of the vector, and update the sampling points by combining the distribution length and the distribution vector.

[0198] The distribution length refers to the offset distance of each sampling point along the normal direction. The distribution length can be obtained by taking the quotient of the total vector length and the number of points. Starting from each boundary point, sampling points are equidistantly arranged along the corresponding distribution vector direction according to the distribution length, which is the updated sampling point.

[0199] The distribution vector is determined by the curvature and normal direction of the boundary points in the high-density area, and the distribution length is calculated by adjusting the spacing and the number of points. This allows the sampling points to be updated adaptively along the boundary normal direction, enabling the sampling points to more accurately fit the edge projection features of the covering material, thereby improving the accuracy of quality inspection.

[0200] Based on the same inventive concept, embodiments of the present invention provide a quality inspection system based on machine vision, employing the following technical solution: The acquisition module is used to acquire a set of images; A memory for storing a program of any one of the machine vision-based quality inspection methods; The processor is the unit of memory that allows programs to be loaded and executed by the processor.

[0201] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0202] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A quality inspection method based on machine vision, characterized in that, include: Step 100: Acquire a set of images, and generate and send a preset light illumination command based on the set of images; Step 101: In response to the light illumination command, determine the projection data of the projection receiving surface, and combine the projection data with the preset sampling points to determine the point image; Step 102: Determine the curve of light intensity change over time at the point based on the image of the point; Step 103: Determine the calculation curve according to the change curve and the light illumination command; Step 104: Determine the valley time based on the calculation curve, and determine the shading period based on the valley time; Step 105: Determine the time deviation based on the occlusion period and the preset theoretical period, and determine the cumulative duration based on the time deviation; Step 106: Determine the frequency domain of the change based on the change curve, and take the maximum frequency of the frequency domain of the change as the oscillation frequency; Step 107: Determine the frequency deviation based on the oscillation frequency and the preset working frequency, and determine the mass deviation by combining the frequency deviation and the cumulative duration; Step 108: When the quality deviation is greater than a preset deviation threshold, a coating quality warning is generated in response to the quality deviation.

2. The quality inspection method based on machine vision according to claim 1, characterized in that, The method for correcting the coating quality warning also includes: Step 109: When the quality deviation is not greater than a preset deviation threshold, determine the shadow region based on the image set; Step 110: Determine the shadow outline based on the shadow area; Step 111: Determine the shadow width based on the shadow outline, and determine the width variation based on the shadow width; Step 112: When the width change is greater than a preset change threshold, determine the torsion coefficient according to the width change; Step 113: Determine the edge straightness and edge parallelism of the contour based on the shadow contour, and determine the edge deviation by combining the edge straightness and edge parallelism; Step 114: When the edge deviation is greater than the preset edge threshold, determine the offset coefficient by combining the edge straightness and edge parallelism; Step 115: When the edge deviation is greater than the preset edge threshold and the width change is greater than the preset change threshold, the deformation coefficient is determined according to the torsion coefficient and the offset coefficient, and a covering quality warning is generated based on the deformation coefficient.

3. The quality inspection method based on machine vision according to claim 2, characterized in that, It also includes a method for determining the tension coefficient, the method for determining the tension coefficient comprising: Step 200: When the width change is not greater than a preset change threshold, determine the irradiation direction based on the light irradiation command, and determine the directional angle based on the irradiation direction and the covering direction; Step 201: Determine the number of partitions based on the included directional angle; Step 202: Determine the shadow sub-regions by combining the shadow area and the number of partitions, and determine the sub-variation curves based on the shadow sub-regions; Step 203: Determine the sub-mass deviation based on the sub-variation curve; Step 204: Arrange the sub-mass deviations according to the positional order of the shaded sub-regions to form a deviation distribution sequence; Step 205: Determine the deviation gradient based on the deviation distribution sequence; Step 206: When the deviation gradient is greater than a preset gradient threshold, determine the wrapping tension coefficient based on the deviation gradient; Step 207: Generate a coating quality warning based on the tension coefficient.

4. The quality inspection method based on machine vision according to claim 3, characterized in that, The method for determining the sub-mass deviation also includes: Step 208: When the deviation gradient is greater than a preset gradient threshold, determine the candidate position based on the width change; Step 209: Based on the candidate position, retrieve candidate parameters from edge straightness and edge parallelism, and determine the calculation position according to the candidate parameters; Step 210: Determine the frequency domain segment based on the changed frequency domain and the preset time window; Step 211: Determine the instantaneous frequency based on the frequency domain segment, and determine the sequence of instantaneous frequency changes over time according to the instantaneous frequency; Step 212: Determine the trend slope based on the change sequence; Step 213: Retrieve the calculated slope from the trend slope according to the calculated position; Step 214: Determine the flipping feature based on the calculated slope, and determine the flipping duration according to the flipping feature; Step 215: When the flipping time is less than a preset time threshold, determine the joint time based on the flipping time; Step 216: Determine the joint parameters based on the joint time, and update the sub-mass deviation according to the joint parameters.

5. The quality inspection method based on machine vision according to claim 3, characterized in that, It also includes a method for generating a warning about abnormal light transmission, the method comprising: Step 217: When the tension coefficient is greater than the preset tension threshold, determine the sampling density based on the tension coefficient; Step 218: Determine the internal points according to the sampling density and shadow area, and generate an internal light intensity curve based on the internal points; Step 219: Determine the light transmission deviation based on the internal light intensity curve and the preset reference light transmission curve; Step 220: Determine the light transmission deviation sequence based on the light transmission deviation, and determine the normalized weight in combination with the tension coefficient; Step 221: Determine the comprehensive sequence by combining the normalized weights, the light transmission deviation sequence, and the deviation distribution sequence; Step 222: Determine the comprehensive variance based on the comprehensive sequence, and generate a warning of abnormal light transmission in response to the comprehensive variance.

6. The quality inspection method based on machine vision according to claim 5, characterized in that, The method for determining the shaded area also includes: Step 300: Determine the light intensity from the light illumination command, and determine the shadow reproduction degree of each covering material according to the directional angle and the light intensity; Step 301: Determine the final reproduction degree of each wrapping by combining the shadow reproduction degree with the preset roughness of each material, and determine the reference material according to the final reproduction degree; Step 302: Retrieve the reference outline and reference width corresponding to the reference material from the shadow outline and shadow width; Step 303: Determine the spatial pose of the reference material by combining the reference profile and reference width; Step 304: Determine the calculation model according to the spatial pose and the preset mechanism model; Step 305: Determine the simulation area of ​​other coating materials according to the calculation model, and update the shadow area based on the simulation area.

7. The quality inspection method based on machine vision according to claim 6, characterized in that, The method for determining the sampling point location also includes: Step 306: Determine the theoretical light intensity based on the sampling points and calculation model, and determine the actual light intensity based on the shadow area; Step 307: Determine the light intensity ratio by combining the theoretical light intensity and the actual light intensity, and determine the reliability of the location based on the light intensity ratio; Step 308: When the confidence level of the point is greater than the preset untrusted threshold, retrieve the point to be adjusted from the sampled points based on the confidence level of the point. Step 309: Determine the spatial coordinates of the point to be adjusted in the covering material by combining the point to be adjusted and the calculation model, and determine the offset vector by combining the spatial coordinates and the calculation model; Step 310: Update the sampling point position according to the offset vector.

8. The quality inspection method based on machine vision according to claim 7, characterized in that, The method for determining the sampling point location also includes: Step 311: When the confidence level of the point is not greater than the preset unconfidential threshold, determine the spatial surface of the coating material based on the calculation model; Step 312: Determine the set of intersection points between the light rays and the covering material based on the spatial curved surface and the irradiation direction; Step 313: Determine the light density distribution on the projection receiving surface by combining the set of intersection points and the calculation model; Step 314: Determine the high-density region by combining the light density distribution and the preset density threshold; Step 315: Determine the reduced adjustment spacing between each sampling point according to the high-density area; Step 316: Update the sampling points according to the adjusted spacing.

9. A quality inspection method based on machine vision according to claim 8, characterized in that, The method for determining the sampling point location also includes: Step 317: Determine the region boundary based on the high-density region, and determine the boundary points according to the region boundary; Step 318: Determine the curvature and normal direction of the boundary points according to the boundary points, and determine the vector length based on the curvature of the boundary points; Step 319: Determine the distribution vector by combining the normal direction and the vector length; Step 320: Determine the number of points based on the high-density area and the adjusted spacing, and determine the total length of the vector based on the vector length; Step 321: Determine the distribution length by combining the number of points and the total length of the vector, and update the sampling points by combining the distribution length and the distribution vector.

10. A quality inspection system based on machine vision, characterized in that, include: The acquisition module is used to acquire a set of images; A memory for storing a program of a machine vision-based quality inspection method as described in any one of claims 1 to 9; The processor is the unit of memory that allows programs to be loaded and executed by the processor.

Images