A visual auxiliary-based electronic precision component burr detection method and system

Abstract

The embodiment of the present application provides a kind of visual auxiliary electronic precision component burr detection method and system, the method comprises: obtaining the surface image sequence and the gray mutation value in surface image sequence of electronic precision component corresponding under different illumination conditions;Determine suspected burr area based on gray mutation value;Obtain the gray scale change difference value of suspected burr area and the illumination symmetry area corresponding to suspected burr area under different illumination conditions, and determine target burr area based on gray scale change difference value;Determine the light and dark features of target burr area and the adaptive gray threshold of target burr area;According to light and dark features and adaptive gray threshold, the image segmentation of local image is carried out to the image to be segmented where target burr area is located, and target burr segmentation image is obtained.The embodiment of the present application can carry out secondary detection according to gray mutation value and gray scale change difference value, and multiple judgment is carried out to burr area, and then the accuracy of identification is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to a visual auxiliary electronic precision component burr detection method and system. BACKGROUND

[0002] The surface burr of the precision electronic structure is usually related to the manufacturing process. Abnormalities occur in the process of removing or imaging the surface material, causing tiny burrs on the surface. These tiny burrs have a serious impact on the performance of the precision electronic structure, such as causing short circuits between structures, electrical leakage, uneven mechanical stress distribution, etc. Accurate identification of these surface burrs can effectively prevent the production of defective products and has important significance in early warning of equipment, etc.

[0003] Since the surface burr is usually generated with a large number of random factors, such as microscopic unevenness of the material and microscopic fluctuations of the process parameters, etc., all of which will cause random distribution of the surface burr of the electronic structure. Different angles of the burr will cause differences in local image gray scale, etc. In the existing image recognition technology, a single image threshold or a single light angle condition is usually used for recognition. Different burrs cause different light reflections, which reduces the accuracy of part of the burr recognition. SUMMARY

[0004] The electronic precision component burr detection method and system under visual assistance provided by the embodiments of the present application can solve the problem of low accuracy of burr recognition in the prior art.

[0005] In one aspect, the embodiments of the present application disclose an electronic precision component burr detection method under visual assistance, which comprises:

[0006] Obtaining a surface image sequence of an electronic precision component under different light conditions and a gray scale mutation value in the surface image sequence; the gray scale mutation value is used to represent the degree of gray scale mutation along adjacent sections of a virtual emission line;

[0007] Determining a suspected burr area based on the gray scale mutation value;

[0008] Obtaining a gray scale change difference value of the suspected burr area under the different light conditions and a light symmetry area corresponding to the suspected burr area, and determining a target burr area based on the gray scale change difference value; the light symmetry area is symmetric about the light source center line with respect to the suspected burr area;

[0009] Determining the light and dark features of the target burr area and the adaptive gray scale threshold of the target burr area;

[0010] Based on the brightness and darkness features and the adaptive grayscale threshold, local image segmentation is performed on the image to be segmented where the target burr region is located, to obtain the target burr segmentation image.

[0011] Optionally, determining the suspected spurious region based on the grayscale abrupt change value includes:

[0012] For any image in the surface image sequence, each virtual emission line is divided into multiple equal-length sub-segments, and the average gray value of each sub-segment is calculated;

[0013] Based on the average gray value of each sub-segment, determine the gray value abrupt change between adjacent segments;

[0014] Based on the grayscale abrupt change value, the probability of suspected burrs is determined;

[0015] Based on the probability of suspected burrs and the first preset threshold, the suspected burr region is determined.

[0016] Optionally, obtaining the difference in grayscale variation between the suspected burr region and its symmetrical region under different lighting conditions, and determining the target burr region based on the difference in grayscale variation, includes:

[0017] For each suspected burr area, determine the illumination symmetry region of the suspected burr area with respect to the center line of the light source;

[0018] Under different lighting conditions, obtain the average gray value sequence corresponding to the suspected burr region and its lighting symmetrical region.

[0019] Based on the average gray value sequence, determine the confidence probability of the suspected spurious region;

[0020] The target burr region is determined based on the confidence probability and the second preset threshold.

[0021] Optionally, determining the light and dark features of the target burr region includes:

[0022] Calculate the first average gray value of a first predetermined range centered on the target burr region; the first predetermined range includes the target burr region and is a closed region whose area is greater than the area of ​​the target burr region.

[0023] Calculate the second average gray value of a second predetermined range that is entirely within the first predetermined range and smaller than the first predetermined range;

[0024] The light and dark features of the target burr region are determined based on the first average gray value and the second average gray value.

[0025] Optionally, determining the light and dark features of the target burr region based on the first average gray value and the second average gray value includes:

[0026] If the first average gray value is less than the second average gray value, the light and dark features corresponding to the target burr area are determined to be highlight features;

[0027] If the first average gray value is greater than or equal to the second average gray value, the light and dark features corresponding to the target burr region are determined to be low dark features.

[0028] Optionally, determining the adaptive grayscale threshold for the target burr region includes:

[0029] Obtain the image region corresponding to the first predetermined range;

[0030] The adaptive grayscale threshold is determined based on the grayscale distribution of the image region.

[0031] Optionally, before determining the light and dark features of the target burr region and the adaptive grayscale threshold of the target burr region, the method further includes:

[0032] For each image in the surface image sequence, a grayscale difference metric between the target burr region and the illumination symmetry region in the image is determined; the grayscale difference metric is used to characterize the absolute difference between the average grayscale values ​​of the target burr region and the illumination symmetry region in the same surface image.

[0033] The image to be segmented is determined from the surface image sequence based on the grayscale difference metric.

[0034] Optionally, the method further includes:

[0035] When the target burr region is the highlighted feature, pixels in the image to be segmented whose grayscale value is greater than or equal to the adaptive grayscale threshold are marked as burr pixels.

[0036] When the target spiky region is the low-dark feature, pixels in the image to be segmented whose gray value is less than the adaptive gray value threshold are marked as spiky pixels.

[0037] Based on the burr pixels, the target burr segmentation image is obtained.

[0038] Optionally, the method further includes:

[0039] Isolated noise pixels in the target spur segmentation image are removed; the isolated noise pixel is a single pixel whose gray value is inconsistent with the pixel gray value within a preset neighborhood range.

[0040] The denoised target burr segmentation image is subjected to edge smoothing processing, and the number, area and distribution location information of the target burrs are counted to generate a detection report.

[0041] On the other hand, embodiments of the present invention disclose a vision-assisted burr detection system for precision electronic components, the system comprising:

[0042] The acquisition module is used to acquire surface image sequences of electronic precision components under different lighting conditions and grayscale abrupt change values ​​in the surface image sequences; the grayscale abrupt change values ​​are used to characterize the degree of grayscale abrupt change along adjacent segments of the virtual emission line;

[0043] The determination module is used to determine suspected burr regions based on the grayscale abrupt change values;

[0044] The acquisition unit is further configured to acquire the difference in grayscale change between the suspected burr area and the corresponding symmetrical region under different lighting conditions;

[0045] The determining unit is further configured to determine the target burr region based on the grayscale change difference value; the illumination symmetry region and the suspected burr region are symmetrical about the center line of the light source; and determine the brightness characteristics of the target burr region and the adaptive grayscale threshold of the target burr region.

[0046] The module is used to perform local image segmentation on the image to be segmented where the target burr region is located based on the brightness and darkness features and the adaptive grayscale threshold, so as to obtain the target burr segmentation image.

[0047] In another aspect, embodiments of the present invention also disclose an electronic device, the electronic device including a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors as described above in the vision-assisted method for detecting burrs on precision electronic components.

[0048] This invention also discloses a readable storage medium, which, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to perform the aforementioned vision-assisted method for detecting burrs on precision electronic components.

[0049] The embodiments of the present invention have the following advantages:

[0050] In the process of burr detection for precision electronic components, this invention first acquires surface image sequences and grayscale abrupt change values ​​in the surface image sequences corresponding to the precision electronic components under different lighting conditions. The grayscale abrupt change values ​​characterize the degree of grayscale abrupt change along adjacent segments of a virtual emission line. Based on the grayscale abrupt change values, suspected burr regions are identified. The difference in grayscale changes between the suspected burr regions and their corresponding symmetrical regions under different lighting conditions is acquired, and the target burr region is determined based on this difference in grayscale changes. The symmetrical regions and the suspected burr regions are symmetrical about the center line of the light source. The brightness and darkness features of the target burr region and its adaptive grayscale threshold are determined. Local image segmentation is performed on the image to be segmented based on the brightness and darkness features and the adaptive grayscale threshold to obtain a segmented image of the target burr. This invention can perform a first detection based on the grayscale abrupt change values ​​to identify suspected burr regions, and then perform a second detection based on the grayscale change difference values ​​to determine the authenticity of the suspected burr regions, thereby improving the accuracy of identification. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating the steps of a vision-assisted method for detecting burrs in precision electronic components according to the present invention. Figure 1 ;

[0053] Figure 2 This is a flowchart illustrating the steps of a vision-assisted method for detecting burrs in precision electronic components according to the present invention. Figure 2 ;

[0054] Figure 3 This is a flowchart illustrating the steps of a vision-assisted method for detecting burrs in precision electronic components according to the present invention. Figure 3 ;

[0055] Figure 4 This invention relates to an image acquisition device;

[0056] Figure 5 This is a schematic diagram of light reflection from burrs according to the present invention;

[0057] Figure 6 This is a schematic diagram of a ring structure according to the present invention;

[0058] Figure 7This is a schematic diagram of the angle of a virtual light emission line according to the present invention;

[0059] Figure 8 This is a schematic diagram of the position of the illuminated symmetrical region of a suspected burr area according to the present invention;

[0060] Figure 9 This is a structural block diagram of a vision-assisted burr detection system for precision electronic components according to the present invention.

[0061] Figure 10 This is a structural block diagram of an electronic device according to the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Reference Figure 1 The flowchart illustrates the steps of an embodiment of a vision-assisted method for detecting burrs in precision electronic components according to the present invention. Figure 1 The method may specifically include the following steps:

[0064] Step S101: Obtain the surface image sequence and grayscale abrupt change value of the electronic precision component under different illumination conditions; the grayscale abrupt change value is used to characterize the degree of grayscale abrupt change of adjacent segments along the virtual emission line;

[0065] Step S102: Based on the grayscale mutation value, identify the suspected burr region;

[0066] Step S103: Obtain the grayscale difference value between the suspected burr area and the corresponding symmetrical region under different lighting conditions, and determine the target burr area based on the grayscale difference value; the symmetrical region and the suspected burr area are symmetrical about the center line of the light source.

[0067] Step S104: Determine the light and dark features of the target burr region and the adaptive grayscale threshold of the target burr region;

[0068] Step S105: Perform local image segmentation on the image to be segmented based on the brightness and darkness features and the adaptive grayscale threshold to obtain the target burr segmentation image.

[0069] Obtain surface image sequences and grayscale abrupt change values ​​in the surface image sequences corresponding to electronic precision components under different illumination conditions; the grayscale abrupt change values ​​are used to characterize the degree of grayscale abrupt change in adjacent segments along the virtual emission line.

[0070] Different lighting conditions refer to the changes in the illumination received by the surface of electronic precision components during the image acquisition process by adjusting the spatial position of the light source (such as height and horizontal angle).

[0071] For example, different lighting conditions can be used to illuminate the workpiece surface with a light source at different incident angles, thereby forming a series of images with different shadow distributions and reflection characteristics in the camera. The purpose of setting different lighting conditions is to overcome the defect that some burrs are difficult to image due to their reflective properties under single lighting.

[0072] Precision electronic components refer to miniature electronic elements or assemblies that require high-precision surface quality control during manufacturing, such as integrated circuit boards, micro connectors, and ceramic substrates. Their surfaces typically have regular geometric shapes (such as the circular structure addressed in this invention), and any tiny excess protrusions (burrs) can affect their electrical performance or mechanical assembly.

[0073] A surface image sequence refers to an ordered set of digital images obtained by continuously capturing the surface of the same precision electronic component using an industrial camera under different lighting conditions. Each image in the sequence records the two-dimensional grayscale information of the component surface under specific lighting parameters.

[0074] Gray-scale abrupt change values ​​are used to characterize the drastic degree of gray-scale changes in local areas of an image. Specifically, the process of determining gray-scale abrupt change values ​​involves calculating the absolute value (or the normalized difference) of the average gray-scale difference between adjacent equal-length segments along a virtual emission line emanating from the center of the circle. The larger the gray-scale abrupt change value, the higher the probability that there is a gray-scale discontinuity caused by abnormal protrusions or depressions such as burrs at that location. Gray-scale abrupt change values ​​are feature values ​​extracted from image data and used to initially locate potential defects.

[0075] Virtual emission lines refer to a set of straight paths radiating outwards from the geometric center (center) of a circular electronic component image, artificially defined for structured analysis during digital image processing. These lines do not actually exist in the image but serve as paths for the algorithm to scan and analyze grayscale distribution. Their angular intervals are set based on the statistical dimensions of historical spurs to ensure that the entire detection area is covered.

[0076] Adjacent segments refer to continuous small segments divided along a single virtual transmission line according to a predetermined length (usually related to the typical glitch size). Each segment contains several consecutive pixels; the algorithm calculates gray-level abrupt changes by comparing the average gray-level values ​​of adjacent segments, thereby discretizing the continuous gray-level changes along the entire transmission line into a series of comparable local difference signals.

[0077] Based on the grayscale mutation value, suspected spiking areas are identified.

[0078] Among them, the suspected burr area refers to the local coordinate range of the image that is initially judged to have the possibility of burrs in the surface image of electronic precision components based on the degree of gray-scale change obtained by analysis along the virtual emission line, after quantitative calculation and comparison with a preset threshold. This area is not the final detection conclusion, but a candidate target to be confirmed in the subsequent illumination symmetry verification process.

[0079] It should be noted that the grayscale abrupt change value is used in the initial screening and localization stage (first detection). Within a single image, along an adjacent image segment on a virtual emission line (for a single image, extract the average grayscale of all pixels in each small segment on the virtual emission line), the absolute difference in average grayscale between two segments at adjacent spatial locations on the same line is then calculated. The abrupt change intensity of this segment is related to the grayscale of the segments before and after it. The physical meaning is that it can measure the degree of abruptness of a local point relative to its background in the spatial dimension. The larger the value, the more likely the grayscale jump is caused by a protrusion (spiking). The grayscale abrupt change value is used to initially discover suspicious points (suspected spiking areas), but it may contain false alarms (such as textures, dust).

[0080] The grayscale difference between the suspected burr area and the corresponding symmetrical region under different lighting conditions is obtained, and the target burr area is determined based on the grayscale difference. The symmetrical region and the suspected burr area are symmetrical about the center line of the light source.

[0081] The symmetrical region corresponding to the suspected burr area refers to the region on the image plane that is geometrically symmetrical to the suspected burr area about the center line of the light source. The center line of the light source is the projection line of the line connecting the center of the light source and the center of the workpiece onto the image plane, serving as the axis of symmetry in the illumination direction. Due to the symmetry, this region is physically equidistant from the center of the workpiece, but is located on the opposite side of the light source.

[0082] The grayscale variation difference value is used to characterize the degree of inconsistency between the average grayscale value change trends of suspected spurious areas and symmetrically lit areas under multiple different lighting conditions. Specifically, it can be obtained by calculating the statistical characteristics (such as the sum of absolute differences, variance, etc.) of the grayscale difference sequence of the two areas in images under different lighting conditions and then normalizing them. The larger the value, the more inconsistent the responses of the two areas to changes in lighting.

[0083] For example, the grayscale variation value measures the degree of inconsistency between the grayscale values ​​of the suspected burr area and the illuminated symmetrical area during the variation process.

[0084] It should be noted that the grayscale variation difference value is used in the secondary determination stage (secondary verification). For multiple image sequences, between the same suspected spur area and its symmetrical area, the average grayscale of all pixels in the two undetermined areas (suspected area and symmetrical area) is extracted for multiple images (under different lighting conditions). The absolute difference of the average grayscale of the two areas at different spatial locations under the same lighting is calculated and accumulated across lighting conditions. This measures the degree of inconsistency in the response of the two symmetrical points to changes in lighting conditions in terms of time or conditions. The larger the value, the more different the physical properties of the two points are (one is a spur, and the other is a smooth surface). This is used to verify the authenticity of suspicious points and eliminate false alarms caused by normal surface texture or uneven lighting. The calculation is the difference in the average grayscale of the two pixel areas under multiple conditions.

[0085] The light source centerline refers to a key virtual reference axis in the three-dimensional spatial model of a vision inspection system. It is defined as a straight line passing through the optical center of the adjustable light source, the center of the circular electronic component image (obtained by Hough transform), and the optical center of the industrial camera lens (or a related reference point). This axis is the axis of symmetry of the entire lighting system and the spatial geometric reference for all symmetric operations of illumination.

[0086] The target burr area refers to the image area that, after a secondary verification process, is selected from the suspected burr areas and ultimately determined to have a genuine burr defect. The determination is based on the fact that the difference in grayscale variation between the suspected area and its illuminated symmetrical area exceeds a high preset threshold, indicating that its physical properties (protruding burrs) are fundamentally different from the smooth surface properties of the symmetrical area.

[0087] Determine the light and dark features of the target burr region and the adaptive grayscale threshold of the target burr region.

[0088] The brightness and darkness characteristics of the target burr area refer to the brightness attributes exhibited by the target burr area under specific lighting conditions. Due to the relationship between the spatial orientation of the burr and the incident direction of light, some burrs appear as bright areas (light reflected into the camera) and some appear as dark areas (light is blocked or scattered). The determination of brightness and darkness characteristics depends on the trend of the change in the average gray value of the local area.

[0089] The adaptive grayscale threshold for the target burr region refers to the grayscale threshold used for image binarization. This threshold is determined based on the difference in grayscale histograms between the foreground (burrs) and the background (normal surface) in the enlarged local image. The grayscale value that maximizes the difference in the number of pixels between the two is selected as the segmentation threshold, thereby enabling personalized segmentation of burrs with different brightness and darkness features.

[0090] Based on the brightness and darkness features and the adaptive grayscale threshold, local image segmentation is performed on the image to be segmented to obtain the target burr segmentation image.

[0091] The image to be segmented, where the target burr region is located, refers to the image with the largest gray-level difference between it and the illuminated symmetrical region selected from the surface image sequence, which is taken as the best representation image corresponding to the target burr region.

[0092] Local image segmentation refers to pixel-level partitioning of the target spur region in an image to be segmented, used to separate the spur region from the background region. For example, adaptive grayscale thresholding can be used for segmentation, and the segmentation direction can be determined based on the brightness characteristics of the spur, ultimately generating a binarized spur segmentation image.

[0093] A target burr segmentation image refers to a binary image obtained after local image segmentation, where burr areas are clearly marked with white or black pixels, while background areas are represented by the opposite color. This image clearly reflects the shape, location, and contour information of the burrs and can be directly used for quality inspection, defect report generation, or subsequent processing.

[0094] For example, the threshold segmentation direction is determined based on the brightness and darkness features, and then the image is binarized using an adaptive grayscale threshold to separate the burr region from the background, generating a binarized target burr segmentation image. This can achieve personalized segmentation of burrs with different brightness and darkness features, ensuring the accuracy and consistency of the segmentation results.

[0095] In the process of burr detection for precision electronic components provided by this invention, the following steps are first taken: First, surface image sequences and grayscale abrupt change values ​​of the surface image sequences corresponding to the precision electronic components under different lighting conditions are acquired. These grayscale abrupt change values ​​characterize the degree of grayscale abrupt change along adjacent segments of a virtual emission line. Based on these grayscale abrupt change values, suspected burr regions are identified. Then, the grayscale change difference values ​​between the suspected burr regions and the corresponding symmetrical regions under different lighting conditions are obtained, and the target burr region is determined based on these grayscale change difference values. The symmetrical regions and the suspected burr regions are symmetrical about the center line of the light source. The brightness and darkness features of the target burr region and its adaptive grayscale threshold are determined. Based on the brightness and darkness features and the adaptive grayscale threshold, local image segmentation is performed on the image to be segmented where the target burr region is located, resulting in a segmented image of the target burr. This invention allows for a first detection based on grayscale abrupt change values ​​to identify suspected burr regions, followed by a second detection based on grayscale change difference values ​​to determine the authenticity of the suspected burr regions, thereby improving the accuracy of identification.

[0096] In an optional embodiment of the present invention, determining the suspected burr region based on grayscale abrupt change values ​​may specifically include the following steps:

[0097] Step S1021: For any image in the surface image sequence, divide each virtual emission line into multiple equal-length sub-segments and calculate the average gray value of each sub-segment;

[0098] Step S1022: Determine the grayscale abrupt change value between adjacent segments based on the average grayscale value of each sub-segment;

[0099] Step S1023: Determine the probability of suspected spikes based on the grayscale mutation value;

[0100] Step S1024: Based on the probability of suspected punctures and the first preset threshold, determine the suspected puncture area.

[0101] For any image in the surface image sequence, each virtual emission line is divided into multiple segments of equal length, and the average gray value of each segment is calculated.

[0102] In this context, equal-length sub-segments refer to line segment units that uniformly divide each virtual emission line according to a fixed pixel length or physical length. The length of each sub-segment is determined based on the average size of burrs in historical detection data, ensuring that burr features can be effectively captured within at least one sub-segment, thereby achieving refined grayscale sampling along the emission line.

[0103] A virtual emission line refers to a virtual straight line path that starts from the center of the detected image and extends radially outward at preset minimum angular intervals.

[0104] The average gray value of each sub-segment refers to the value obtained by taking the arithmetic mean of the gray values ​​of all pixels within a certain sub-segment of an image. This average gray value is used to characterize the overall brightness level of that sub-segment region under specific lighting conditions.

[0105] It should be noted that by uniformly dividing each virtual emission line into multiple equal-length sub-segments and calculating the average gray value of pixels within each sub-segment region to form a gray-scale distribution sequence along the emission line, this process achieves systematic gray-scale feature extraction of the surface of the structural component in the radial direction.

[0106] The grayscale abrupt change value between adjacent segments is determined based on the average grayscale value of each sub-segment.

[0107] Adjacent segments refer to two sub-segments that are adjacent to each other on a single virtual transmission line. Since the virtual transmission line radiates outward from the center, adjacent segments usually refer to the pairing relationship formed by a sub-segment closer to the center (front segment) and a sub-segment farther away from the center (back segment) in the radial direction.

[0108] The grayscale abrupt change value is used to characterize the degree of grayscale change between adjacent segments. This grayscale abrupt change value is determined based on the absolute value of the difference between the average grayscale values ​​of two adjacent segments, or based on the grayscale difference between the current segment and the two segments before and after it. The larger the value, the higher the probability that there is a sudden grayscale change caused by spikes at this boundary.

[0109] It should be noted that each segment divided on the virtual transmission line is treated as a processing unit, and its average gray value is calculated to characterize the overall brightness at that location. Then, the average gray values ​​of adjacent segments in the radial direction are compared, and a quantized gray value change is obtained by calculating the absolute value of the difference or a more complex combination of the differences.

[0110] It should be noted that on a smooth surface, the average gray value of adjacent segments should transition smoothly with small abrupt changes; however, when the path passes through a protruding or recessed burr, its edge will cause a sudden increase or decrease in gray value, resulting in a significant peak in gray value abrupt change.

[0111] The probability of suspected spikes is determined based on the grayscale abrupt change value.

[0112] The probability of a suspected glitch refers to the quantified value of the likelihood that a certain sub-region is a glitch, calculated based on grayscale mutation values. It is usually normalized and ranges from 0 to 1. This probability value is used for preliminary identification and screening of potential glitch regions.

[0113] Based on the probability of suspected punctures and a first preset threshold, the suspected puncture region is determined.

[0114] The first preset threshold refers to a critical value set for judging the probability of suspected spikes during the initial screening stage. This threshold is a constant predetermined based on historical experimental data, experience, or desired detection sensitivity.

[0115] In an optional embodiment of the present invention, obtaining the difference in grayscale variation between the suspected burr region and its symmetrical region under different lighting conditions, and determining the target burr region based on the difference in grayscale variation, may specifically include the following steps:

[0116] Step S1031: For each suspected burr area, determine the illumination symmetry area of ​​the suspected burr area with respect to the center line of the light source;

[0117] Step S1032: Obtain the average gray value sequence corresponding to the suspected burr area and its symmetrical area under different lighting conditions;

[0118] Step S1033: Determine the confidence probability of the suspected spurious region based on the average gray value sequence;

[0119] Step S1034: Determine the target burr region based on the confidence probability and the second preset threshold.

[0120] For each suspected burr area, determine the illumination symmetry region of the suspected burr area with respect to the center line of the light source.

[0121] The symmetrical region of illumination refers to an image sub-region that is geometrically mirror-symmetrical to the suspected burr region with respect to the center line of the light source. Theoretically, this region should have similar illumination conditions to the suspected burr region, but its grayscale variation pattern may differ under normal surface conditions.

[0122] Obtain the average gray value sequence corresponding to the suspected burr area and its symmetrical area under different lighting conditions.

[0123] The sequence of average gray values ​​corresponding to the suspected burr area and its corresponding symmetrical illumination area refers to two ordered sets of gray values ​​extracted and arranged from multiple images acquired on the same electronic component surface under different illumination angles or heights, targeting a specific suspected burr area and its corresponding symmetrical illumination area. This sequence records the evolution of the average gray values ​​of the two areas over time or under varying illumination parameters, used for subsequent analysis of the consistency or difference in their responses to changes in illumination, thereby determining whether the suspected area is a genuine burr.

[0124] The confidence probability of suspected spurious regions is determined based on the average gray value sequence.

[0125] The confidence probability of a suspected burr region refers to the quantified value of the likelihood that the region is a real burr, calculated by normalization based on the difference in the average grayscale value sequence between the suspected burr region and its symmetrical region under multi-angle illumination conditions. This probability reflects the degree of inconsistency in the grayscale response patterns of the two regions during illumination changes. The higher the inconsistency, the greater the confidence probability, indicating that the region is more likely to be a real burr; conversely, it is more likely to be a misjudged normal surface region.

[0126] The target burr region is determined based on the confidence probability and the second preset threshold.

[0127] The second preset threshold refers to the critical probability value used in the secondary confirmation stage to determine whether a suspected burr area is a real burr. This threshold is usually set based on experimental experience or statistical learning. If the confidence probability of a suspected burr area calculated through grayscale difference analysis is higher than this threshold, the system determines that the area is a real burr; otherwise, it is considered a false positive and excluded. The setting of the second preset threshold directly affects the balance between detection accuracy and false alarm rate.

[0128] The target spur region refers to the local image region that is ultimately identified as a real spur after initial screening via grayscale abrupt changes and secondary confirmation via illumination symmetry. This region has undergone two screenings, and its location and extent have been determined in the optimally illuminated image for subsequent adaptive segmentation.

[0129] In an optional embodiment of the present invention, determining the light and dark features of the target burr region may specifically include the following steps:

[0130] Step S1041: Calculate the first average gray value of a first predetermined range centered on the target burr region; the first predetermined range is a closed region that includes the target burr region and whose area is greater than the area of ​​the target burr region.

[0131] Step S1042: Calculate the second average gray value of the second predetermined range that is completely within the first predetermined range and is smaller than the first predetermined range;

[0132] Step S1043: Determine the light and dark features of the target burr area based on the first average gray value and the second average gray value.

[0133] Calculate the first average gray value of a first predetermined range centered on the target burr region; the first predetermined range is a closed region that includes the target burr region and whose area is greater than the area of ​​the target burr region.

[0134] Here, the first predetermined range centered on the target burr region refers to an extended region defined around the target burr region in the image. This region is typically larger than the original burr region, such as a rectangular or circular region obtained by extending the boundary of the burr region outward by a fixed pixel distance or by scaling it up proportionally. The first predetermined range is used to characterize the background pixels that include the burr and its surrounding area.

[0135] The first average gray value refers to the average gray value of all pixels within a first predetermined range. This value reflects the overall brightness level, including the burrs and their surrounding background, and is compared with the average gray value subsequently calculated from a smaller range (such as a second predetermined range) to determine the brightness characteristics of the burrs.

[0136] Calculate the second average gray value of a second predetermined range that is completely within the first predetermined range and smaller than the first predetermined range.

[0137] The second predetermined range refers to a local sub-region that further shrinks inward from the first predetermined range centered on the target burr region. This region is typically determined by proportionally reducing the side length of the first predetermined range or by adaptively adjusting it based on the burr size. The second predetermined range focuses on the core burr region and its immediate vicinity, reducing interference from surrounding background pixels and thus more accurately extracting the grayscale distribution features of the burr itself.

[0138] The second average gray value refers to the average gray value of all pixels within the second predetermined range. This value reflects the local brightness level of the glitch core and its neighboring areas. By comparing it with the first average gray value, the gray value change trend from the outside to the inside can be analyzed, thereby determining the brightness characteristics of the glitch.

[0139] The light and dark features of the target burr region are determined based on the first average gray value and the second average gray value.

[0140] The brightness and darkness characteristics of the target burr area refer to the brightness attribute category exhibited by the area under specific lighting conditions, that is, a qualitative description of whether the burr appears bright or dark relative to the surrounding background surface. This characteristic is determined by comparing the average gray value change trend of different range areas (such as the first predetermined range and the second predetermined range), reflecting the reflection or occlusion effect caused by the spatial orientation of the burr and the interaction with light.

[0141] In an optional embodiment of the present invention, determining the light and dark features of the target burr region based on the first average gray value and the second average gray value may specifically include the following steps:

[0142] Step S10431: When the first average gray value is less than the second average gray value, determine the light and dark features corresponding to the target burr area as the highlight features.

[0143] Step S10432: If the first average gray value is greater than or equal to the second average gray value, determine the light and dark features corresponding to the target burr area as low dark features.

[0144] For example, when determining the brightness characteristics of the target burr region, the system compares a first average gray value (representing the overall brightness of a larger area including the burr and background) with a second average gray value (representing the local brightness of the burr core and adjacent areas): if the first average gray value is greater than or equal to the second average gray value, it indicates that the brightness decreases from the periphery to the core, and the burr region is determined to have a low-dark characteristic; conversely, if the first average gray value is less than the second average gray value, it indicates that the brightness increases inward, and the burr region is determined to have a high-brightness characteristic.

[0145] In an optional embodiment of the present invention, determining the adaptive grayscale threshold of the target burr region may specifically include the following steps:

[0146] Step S1044: Obtain the image region corresponding to the first predetermined range;

[0147] Step S1045: Determine the adaptive grayscale threshold based on the grayscale distribution of the image region.

[0148] Obtain the image region corresponding to the first predetermined range.

[0149] The image region corresponding to the first predetermined range refers to a local sub-image range defined in the original detection image, centered on the target burr region and expanded according to a preset size. This region not only includes the target burr itself, but also covers a certain range of background pixels around it, providing overall grayscale context information about the burr and the surrounding surface.

[0150] An adaptive grayscale threshold is determined based on the grayscale distribution of the image region.

[0151] The grayscale distribution of an image region refers to the statistical characteristics of the grayscale values ​​of all pixels within that region, typically represented by a grayscale histogram. This distribution reflects the number or frequency of pixels at different grayscale levels within the region, describing the region's brightness concentration trend, dispersion, and the presence of multi-peak characteristics.

[0152] An adaptive grayscale threshold is a grayscale critical value dynamically calculated using a specific algorithm (such as the maximum inter-class variance method or the maximum difference method of grayscale histogram) based on the grayscale distribution characteristics of the current image region. This threshold is used for binary segmentation of the image region, dividing pixels into foreground (such as spiking) and background categories, and can adaptively adjust according to the brightness characteristics and distribution changes of the region to achieve accurate segmentation of local features.

[0153] In an optional embodiment of the present invention, the step of determining the light and dark features of the target burr region and the adaptive grayscale threshold of the target burr region may specifically include the following steps:

[0154] Step S106: For each image in the surface image sequence, determine the grayscale difference measure between the target burr region and the illumination symmetry region in the image; the grayscale difference measure is used to characterize the absolute difference between the average grayscale values ​​of the target burr region and its illumination symmetry region in the same surface image.

[0155] Step S107: Determine the image to be segmented from the surface image sequence based on the grayscale difference metric.

[0156] For each image in the surface image sequence, determine the grayscale difference metric between the target burr region and the illuminated symmetrical region in the image; the grayscale difference metric is used to characterize the absolute difference between the average grayscale values ​​of the target burr region and its illuminated symmetrical region in the same surface image.

[0157] The grayscale difference metric between the target spur region and the illuminated symmetrical region in an image refers to the absolute value or sum of squares of the difference in average grayscale values ​​between the target spur region and its illuminated symmetrical region in one or more images under different lighting conditions. This grayscale difference metric is used to characterize the degree of difference in brightness response between the two regions under the same lighting or lighting sequence; the greater the difference, the more likely the target region is to be a real spur.

[0158] The image to be segmented is determined from the surface image sequence based on the gray-level difference metric.

[0159] For example, when determining the image to be segmented in the surface image sequence, the system first selects the image with the largest grayscale difference between the real burr region and the symmetrical region under multi-angle illumination conditions as the best representation image of the burr for each confirmed real burr region. This image can most clearly highlight the contrast features between the burr and the background. Then, a local sub-image is extracted from the burr region as the center and within a predetermined range, which is the final image to be segmented for adaptive threshold segmentation and feature extraction.

[0160] In an optional embodiment of the present invention, the vision-assisted burr detection method for electronic precision components may further include the following steps:

[0161] When the target spur region is a highlight feature, pixels in the image to be segmented whose gray value is greater than or equal to the adaptive gray value threshold are marked as spur pixels.

[0162] Highlighting features refer to a brightness attribute of the target burr area, characterized by a grayscale value that is significantly higher than that of the surrounding normal planar area. This is caused by the outward direction of the burrs being horizontal to the direction of light illumination, resulting in a large amount of light being reflected into the industrial camera, causing the burr area to appear as a highlight in the image.

[0163] A burr pixel refers to a pixel in the image to be segmented that is determined to belong to the target burr region. It is the basic unit that constitutes the outline of the burr image. After being marked, a complete burr segmentation image can be formed.

[0164] When the target burr region exhibits a high-brightness feature with a gray value higher than the surrounding area, pixels in the image to be segmented whose gray value reaches or exceeds the threshold are identified and marked as pixels constituting burrs, thus achieving accurate segmentation of the high-brightness burr feature.

[0165] When the target spur region has low-dark features, pixels in the image to be segmented whose gray values ​​are less than the adaptive gray threshold are marked as spur pixels.

[0166] The low-darkness feature refers to a brightness attribute of the target spiky area, used to characterize that the gray value of this area is significantly lower than the gray value of the surrounding normal planar area. This is caused by the outward direction of the spiky area being clearly perpendicular to the direction of the light source, significantly blocking the light and resulting in the area appearing as a low-darkness feature in the image.

[0167] When the target burr region exhibits a low-dark feature with a gray value lower than the surrounding area, pixels in the image to be segmented whose gray value is less than the threshold are identified and marked as pixels constituting burrs, thus achieving accurate segmentation of low-dark feature burrs.

[0168] Based on the spur pixels, the target spur segmentation image is obtained.

[0169] After identifying all burr pixels in the image to be segmented that meet the criteria, these pixels are integrated and their contours extracted to obtain an image that accurately reflects the location, shape, and extent of the real burrs, i.e., the target burr segmentation image.

[0170] In an optional embodiment of the present invention, the vision-assisted burr detection method for electronic precision components may further include the following steps:

[0171] Isolated noise pixels in the target spur segmentation image are removed; isolated noise pixels are single pixels whose gray values ​​are inconsistent with those of pixels within a preset neighborhood range.

[0172] In the segmentation image of the target burr, misjudged pixels caused by factors such as segmentation algorithm error and light reflection interference are spatially isolated single pixels. They are usually characterized by grayscale that differs from the surrounding preset neighborhood pixels and do not belong to the components of the real burr, which will interfere with the accurate identification of the burr.

[0173] The preset neighborhood range refers to the local image region (such as a 3×3 neighborhood) centered on the target single pixel, which is set to determine isolated noise pixels. It is a commonly used spatial range standard in image processing to determine whether a pixel is isolated, ensuring that only truly isolated misjudged pixels are removed without destroying the continuous pixel structure of real burrs.

[0174] Pixel grayscale value is a numerical value used to characterize the brightness of a single pixel in an image (usually ranging from 0 to 255; the larger the value, the brighter the pixel, and the smaller the value, the darker the pixel). It is a quantitative basis for distinguishing whether a pixel belongs to glitch or noise.

[0175] It should be noted that, in order to optimize the quality of the target spur segmentation image and eliminate interference from non-real spurs, isolated noise pixels in the image need to be removed to finally obtain the real spur image.

[0176] The edge smoothing process is performed on the denoised target burr segmentation image, and the number, area, and distribution location information of the target burrs are counted to generate a detection report.

[0177] The denoised target burr segmentation image refers to the burr segmentation image after removing isolated noise pixels. This image has eliminated isolated misjudged pixels caused by algorithm errors, lighting interference, etc. during the segmentation process, and only retains the continuous pixel regions corresponding to the real burrs. It is the core image result with clearer contours and more accurate information.

[0178] Edge smoothing refers to the optimization operation for denoised spur-segmented images. Its purpose is to eliminate imperfections such as jagged edges and irregular protrusions caused by pixel dispersion at the spur edges, making the spur contours more coherent and smooth. Its core logic is to adjust the grayscale values ​​of the pixels at the spur edge using image processing algorithms (such as mean filtering and Gaussian filtering) to balance the grayscale differences between adjacent pixels.

[0179] Target burrs refer to the real, prominent, tiny physical defects that exist on the surface of precision electronic components. They are the core object of the inspection task and are represented in the image as a continuous pixel area after confirmation, noise reduction, and smoothing. They correspond to actual defects on the surface of the component that could cause short circuits, leakage, or other problems.

[0180] The number of target burrs refers to the total number of independent target burrs identified in the denoised and smoothed image (each independent burr is a non-connected continuous pixel region).

[0181] The area of ​​a target burr refers to the total number of pixels corresponding to each target burr (which can be converted into the actual physical area of ​​the burr on the component surface through the conversion relationship between the number of pixels and the actual physical size), and is a quantitative indicator for measuring the size of the burr.

[0182] The distribution location information of the target burrs refers to the specific coordinates or relative position description of the target burrs on the surface of electronic precision components (based on the image coordinate system or the geometric features of the component itself, such as the distance from the center of the circular component, the annular area it is located in, etc.), which is used to accurately locate the position of each burr.

[0183] A test report is a standardized document that integrates all test results. It includes basic information about the precision electronic components, test parameters (such as lighting conditions, threshold settings, etc.), and key data such as the number, area, and distribution location of burrs.

[0184] After denoising the target burr segmentation image, edge smoothing is used to optimize the coherence and smoothness of the burr contour. This allows for the accurate counting of the total number of real burrs in the image, the size of each burr, and their distribution location. Finally, these detection data are integrated with relevant parameters to generate a standardized inspection report that can be directly used for product quality assessment.

[0185] In summary, the embodiments of the present invention, in the process of burr detection of electronic precision components, firstly acquire the surface image sequence and gray-level abrupt change values ​​of the electronic precision component under different illumination conditions; the gray-level abrupt change values ​​are used to characterize the degree of gray-level abrupt change in adjacent segments along the virtual emission line; based on the gray-level abrupt change values, suspected burr regions are identified; the gray-level change difference values ​​between the suspected burr region and the corresponding illumination symmetrical region under different illumination conditions are acquired, and the target burr region is determined based on the gray-level change difference values; the illumination symmetrical region and the suspected burr region are symmetrical about the center line of the light source; the brightness and darkness features of the target burr region and the adaptive gray-level threshold of the target burr region are determined; based on the brightness and darkness features and the adaptive gray-level threshold, local image segmentation is performed on the image to be segmented where the target burr region is located to obtain the target burr segmented image. The embodiments of the present invention can perform a first detection based on the gray-level abrupt change values ​​to identify suspected burr regions, and then perform a second detection based on the gray-level change difference values ​​to determine the authenticity of the suspected burr regions, thereby improving the accuracy of identification.

[0186] Reference Figure 2 The flowchart illustrates the steps of an embodiment of a vision-assisted method for detecting burrs in precision electronic components according to the present invention. Figure 2 The method may specifically include the following steps:

[0187] Step S201: Set up an image acquisition device to acquire burr images under multiple lighting conditions (a sequence of surface images corresponding to electronic precision components under different lighting conditions).

[0188] By adjusting the height and angle of the light source, burr images under different lighting conditions can be obtained.

[0189] Step S202: Set the burr detection direction line and obtain the suspected burr area based on the grayscale characteristics of the local area caused by the burr.

[0190] Based on the historical average radius of the burr point and the fan-shaped relationship between its distance from the center of the electronic device, the minimum virtual emission line angle is obtained. Then, virtual lines are emitted from the center outwards at the minimum virtual emission line angle interval, and the gray values ​​of the pixels on the virtual lines are counted. Since the incident direction of light is fixed in an image, the pixel gray values ​​along the flat surface are distributed smoothly, while areas with significant abrupt changes are usually caused by burrs.

[0191] Step S203: Change the lighting conditions and compare the consistency between the suspected burrs and the symmetrical area under the lighting to perform a second confirmation of the burrs.

[0192] In symmetrically illuminated areas, the grayscale changes produced by changes in illumination are generally different; conversely, if the suspected burr area is a normal plane, then in symmetrically illuminated areas, the differences in changes between the two are similar as illumination changes.

[0193] Therefore, under different lighting conditions, if the grayscale changes of the symmetrical areas of the suspected burr area are relatively consistent, then the area is a non-burr area; the more inconsistent they are, the more likely it is a burr area.

[0194] Step S204: Obtain the local spur image (image to be segmented) under optimal lighting conditions and its segmentation threshold (adaptive grayscale threshold).

[0195] It should be noted that for a true burr region, the image with the greatest difference from the symmetrical region is selected as the local burr image. The virtual emission line region on this image is used as the radius of a rectangle with the length of this region as the center of the rectangle. The resulting rectangular region is then used as the local region of the burr. The location of this rectangular region in the image is selected as the local image of the burr.

[0196] The spiky region is the area where the relative gray level changes significantly, and there are usually no continuous spiky regions around it. Therefore, the original corresponding spiky image region is used to crop the image on the original image of the spiky image, with the image size being twice the side length of the original image.

[0197] Furthermore, since the enlarged local spur image after cropping usually has a background area at its outer edge and the original inner area is a foreground area containing spurs, there is usually a significant grayscale difference between the two. And because the enlarged area is significantly larger than the original area, the grayscale with the significantly increased number of pixels is the pixel grayscale of the normal plane. Therefore, the grayscale with the largest difference in the number of pixels in the grayscale histograms of the two can be used as the spur segmentation threshold.

[0198] Step S205: Segment the local burr image and identify the burr pixels.

[0199] It should be noted that because some burrs exhibit a bright characteristic while others exhibit a dim characteristic, it is necessary to determine whether the burrs exhibit a bright or dim characteristic during segmentation. The local area surrounding the burrs is usually a normal planar area. The grayscale of the surrounding local area will be diluted by the grayscale of the burrs. The closer to the core of the burr area, the less the local trend is diluted by the grayscale of the burrs.

[0200] Therefore, when the gray level changes from a larger local area to a smaller local area, if the gray level of the local area gradually decreases, it indicates that the burr is in low brightness, and pixels with a gray level less than the threshold are burr pixels. Conversely, if the gray level increases when changing from a larger local area to a smaller local area, it indicates that the burr is in high brightness, and pixels with a gray level greater than the segmentation threshold are burr pixels.

[0201] The main objective of this invention is to identify the optimal segmentation image of local burrs on the surface of a precision electronic structure; this invention obtains the optimal segmentation image of local burrs on the surface of a precision electronic structure under different illumination angles.

[0202] Reference Figure 3 The flowchart illustrates the steps of an embodiment of a vision-assisted method for detecting burrs in precision electronic components according to the present invention. Figure 3 The method may specifically include the following steps:

[0203] Step S301: Set up a burr detection device for precision electronic structures and collect surface images of precision electronic components using an industrial camera.

[0204] Reference Figure 4 This invention illustrates an image acquisition device that acquires surface images of precision electronic structural components. An industrial camera is positioned directly above the center of a stage, with the electronic structural component placed at the center of the stage. A light source capable of high-speed lifting, lowering, and rotating is located on the side. The light source can fully illuminate the surface of the electronic structure and has a lifting function. The illumination angle is freely adjustable, covering a range of 40° to 70°, and can be expanded according to testing requirements. The stage surface is situated in a recess, ensuring the electronic structural component remains stable on the stage during image capture.

[0205] For example, a light source that can be raised, lowered, and rotated at high speed refers to a light source that rotates along a concentric circular track of a structural component, so that it can illuminate light from all other directions of the shelf, and different light intensities can be obtained by adjusting the pitch angle.

[0206] Step S302: Based on the gray-level abrupt change (gray-level abrupt change value) of the structural component image, determine the suspected burr area, and then combine the changes of the symmetrical area under different lighting conditions to perform secondary burr confirmation (secondary determination based on the difference value of gray-level change); and obtain the local image of the burr under the best lighting conditions (the image to be segmented where the target burr area is located) and its segmentation gray-level threshold (adaptive gray-level threshold).

[0207] Burrs on precision electronic surfaces are typically highly random, with varying outward angles and sizes. Because the light source is on one side, different burrs produce different image representations. Since burrs protrude from the surface of the electronic microstructure, they act like small hills relative to the original plane, directly blocking light emitted from the side. If the outward direction of the burr is clearly perpendicular to the light source, the light is significantly blocked, resulting in noticeable grayscale fluctuations in the burr area on the image. Conversely, when the burr is relatively horizontal, its direction is clearly horizontal to the light source. Due to significant light reflection to other locations or reflection into the camera, the burr area appears as a dark or bright region.

[0208] Reference Figure 5 The figure shows a schematic diagram of light reflection from burrs according to the present invention. In the figure, a large amount of light is refracted away at the burrs (a) and (b), while (c) shows a bright feature.

[0209] Therefore, under any lighting conditions, if a certain area of ​​the image exhibits a significant decrease or increase in grayscale along the direction of light attenuation, this area is usually a suspected burr area. Furthermore, since burrs are inherent static defects on electronic surfaces and do not change their shape due to lighting or other factors, while the incident direction of light significantly affects local grayscale changes, the brightness and darkness characteristics of burrs will also change as the incident direction of light gradually changes from horizontal to vertical.

[0210] Considering the randomness of burr distribution, it is generally believed that no new burrs identical to existing suspected burrs will exist at geometrically symmetrical locations under light illumination. Therefore, as the angle of light illumination changes, the more consistent the grayscale changes at the symmetrical locations are with the changes in existing suspected burrs, the more likely the existing suspected burrs are misjudged areas; conversely, the more inconsistent they are, the more likely the existing suspected burrs are genuine burrs. Furthermore, for other suspected burrs, it is possible to gradually determine whether the existing suspected burrs are indeed genuine burrs.

[0211] Therefore, this step first identifies the initial suspected spurious area by analyzing the grayscale decay trend in the direction of light attenuation based on the initial light angle; secondly, it analyzes the consistency of grayscale decay changes between the suspected spurious area and the symmetrical position of the light illumination by analyzing the changes in the incident angle and height of the light, to determine whether it is a real spur; finally, it determines the local grayscale threshold in the image where the real spurs are most obvious.

[0212] It should be noted that the suspected burr area was obtained based on the wide range of light source angle changes and the abrupt changes in grayscale of the emission lines at different angles.

[0213] Reference Figure 6The diagram shows a ring structure of the present invention. Due to the shape characteristics of the burrs, the direction of light reflection will change. The amount of light entering the camera will increase or decrease significantly compared to the flat surface. This will cause the original uniform change direction of light to show an obvious abrupt change trend. Electronic precision structure patterns usually have obvious geometric features, such as circular electronic structural components. From the center of the circular structure outward, the image grayscale can be divided into multiple different hierarchical structures. Along the center outward, the grayscale difference between the hierarchical structures is usually small.

[0214] When a large grayscale fluctuation occurs in a certain area of ​​the ring structure through the emission line from the center, the area at that location is more likely to be a burr.

[0215] Reference Figure 7 This diagram illustrates a virtual emission line angle according to the present invention. When performing grayscale fluctuation difference analysis gradually outward from the center of a circle, a minimum emission line traversal angle needs to be set to ensure that burrs on the circular structural component far from the center are identified. This angle is typically estimated based on historically identified burrs, ensuring that at least one emission line passes through the burr location. According to geometric relationships, the minimum emission line angle and the burr point radius need to form a sector.

[0216] Therefore, it is necessary to obtain the interval angle of the emission line based on the historical average radius of the burrs and the average radius position of the burrs.

[0217] Therefore, this step first obtains the virtual emission line angles on the surface of the electronic structural component; secondly, it identifies areas where grayscale changes abruptly on these emission lines, with the areas of severe abrupt changes being suspected burrs.

[0218] Within the adjustable height range of the light source, multiple illumination angles are set at equal intervals to acquire multiple surface images of the electronic structural component under different light source angles. The initial number of angles needs to be adaptively adjusted according to the actual size of the structural component to ensure sufficient illumination coverage and complete feature sampling. Generally, the number of angles should not be less than five.

[0219] Using Hough transform circle detection, the area of ​​the circular structural component is identified and marked at the center of the circle; by using the detection history of burrs or by manually measuring multiple burrs, the distance of these burrs from the center of the circle and the radius of the burrs are statistically analyzed.

[0220] Based on the above discussion, the minimum angle of the virtual emission line at the center of the circle, according to the relationship between the historical average radius of the burr point and the fan-shaped relationship formed by its distance from the center, is expressed as follows:

[0221]

[0222] In the formula, The minimum angle (in radians) represents the virtual emission line. This indicates the average radius of the burr points as measured historically or manually. This indicates the average distance from the center of the circle to the location of the burr point as measured historically or manually.

[0223] It should be noted that, in If the value is a non-integer (e.g., 17.5°) and not divisible by 360°, then it is necessary to ensure that all burr locations are covered by at least one virtual emission line, while traversing the entire circumference of the circular electronic structure as evenly as possible.

[0224] For example, the calculated non-integer Round up to the nearest integer (e.g., 17.5° is rounded up to 18°), and generate virtual emission lines at intervals based on the rounded angles. If the angle between the last virtual emission line and the first virtual emission line is less than 360°, add an extra ray to complete the circumference.

[0225] For example, with Virtual lines are emitted from the center outwards at angular intervals, and the grayscale values ​​of pixels on the virtual lines are counted.

[0226] Since the incident direction of light is fixed within an image, the pixel grayscale along a flat surface exhibits a stable distribution, while areas with significant abrupt changes are usually caused by spikes. Therefore, the grayscale abrupt change trend of each locality is calculated on the virtual emission line, mainly based on adjacent... The grayscale difference is determined.

[0227] On each transmission line, The length of the pixel count is divided into stages, each stage being a region. The average grayscale value of the pixels in each stage is recorded, and the difference between each stage and the adjacent stages is calculated (to determine the grayscale abrupt change value between adjacent segments). The larger the difference value, the higher the probability of it being a glitch.

[0228] Therefore, the probability that the k-th region on the i-th emission line is a glitch region (the probability of a suspected glitch) is expressed as follows:

[0229]

[0230] In the formula, This indicates the probability that the k-th region on the i-th virtual transmission line is a glitch region; This represents the average gray value of the k-th region on the i-th virtual transmission line.

[0231] It should be noted that when calculating the first point on the virtual transmission line (k=0), out-of-range index entries will be included. Set to 0 to retain only And multiply by 2, that is .

[0232] Similarly, when calculating the last point (k=N) on the virtual transmission line, out-of-range index entries will be considered. Set to 0 to retain only And multiply by 2, that is .

[0233] Furthermore, for each virtual emission line on each image under different lighting conditions, the probability that different regions on the emission line are burrs is calculated.

[0234] The probability threshold (first preset threshold) is set to 0.7. If the probability that a region on a certain transmission line is a glitch is greater than this threshold, then the region is considered a suspected glitch region.

[0235] It should be noted that, for locations with burrs, the angle of the light source is adjusted, and the changes in burrs at symmetrical locations are analyzed to determine whether they are indeed burrs.

[0236] After identifying the suspected burr areas, since normal planes also reflect light, some normal plane areas may be mistaken for burrs. Therefore, this step requires a secondary confirmation of the suspected burr areas. Because the occurrence of true burrs is random—it's rare for two identical burrs to appear on the same electronic component—and their locations are also relatively random, identical burrs generally won't exist in symmetrically illuminated areas. Furthermore, the grayscale of a burr gradually changes as the angle of light incidence changes. In symmetrically illuminated areas, the grayscale changes with illumination are generally different. Conversely, if the suspected burr area is a normal plane, the differences in changes between the two areas are similar in symmetrically illuminated areas.

[0237] Therefore, under different lighting conditions, if the grayscale changes of the symmetrical areas of the suspected burr area are relatively consistent, then the area is a non-burr area; the more inconsistent they are, the more likely it is a burr area.

[0238] Therefore, by setting the illumination height variation condition, the difference in illumination grayscale variation between the symmetrical illumination area of ​​the suspected burr and the original suspected area is used to determine whether it is a real burr area, and the initial suspected burr areas under multiple angles are obtained. Then, one of the images is randomly selected, and all suspected burr areas are marked in the image.

[0239] Reference Figure 8The diagram shows a schematic of the position of the illumination symmetry region of a suspected burr area according to the present invention. Since the light source is on one side of the structure and the center of the light source is on the line connecting the bracket and the center of the circle, the suspected burr area and its virtual emission line are obtained respectively. The angular symmetry is achieved by using the center line of the light rays. The virtual emission line symmetrical along the center line of the light rays is used to take the area at the same position as the center of the circle as the illumination symmetry region of the suspected burr area.

[0240] It should be noted that region k represents a suspected burr region, and k' represents a region symmetrical to region k in terms of illumination. Both regions are segments of a virtual emission line, not circular regions as shown in the diagram.

[0241] Then, each suspected burr area is grouped with its symmetrically illuminated area, and their respective location areas are recorded. The light source is adjusted from the maximum incident angle and height to the minimum height, with the step size based on the adjustable step size of the light source, to obtain multiple sets of gray values ​​for this combination under different illumination conditions.

[0242] It should be noted that the camera position remains unchanged when the lighting conditions change, so the same area in each image represents the same location of the electronic component.

[0243] It should be noted that when the difference between two regions changes significantly with varying illumination, it indicates a high probability that the original suspected spur region is a true spur; conversely, a smaller difference indicates a lower probability that the original suspected spur region is a true spur region. Therefore, the probability that the k-th suspected spur region is a true spur (the confidence probability of a suspected spur region) is expressed as follows:

[0244]

[0245] In the formula, The probability that the kth suspected region of the i-th virtual emission line is a real glitch is represented by norm(); n represents the number of images captured from multiple angles. , Let represent the average gray values ​​of the k-th suspected region and its symmetrical region of the i-th virtual emission line under the m-th illumination condition, respectively.

[0246] For example, a normalization function can normalize the maximum and minimum values, with values ​​ranging from 0 to 1.

[0247] It should be noted that the incident angle consistency control logic of region k (the suspected burr region) and region k' (its symmetrical illumination region) in this invention includes: the light source can be raised, lowered, and rotated through its own support, and after adjustment, the light emission direction is always along the light centerline, ensuring that k and k' are angularly symmetrically distributed along this centerline. When k and k' are both flat metal surfaces without burrs, since the distance between the two regions and the light source is the same and the surfaces are both horizontal, the incident angles of the light received are exactly the same; at the same time, the industrial camera is located directly above the center of the platform, and the angles at which the reflected light from the two regions enters the camera are also consistent, resulting in relatively consistent grayscale values ​​and patterns of change with illumination angle. If k is a true burr region, even if the incident angles of k and k' are kept consistent through light source adjustment, the blocking and scattering effect of the burr on the light will still change the intensity and direction of its reflected light, causing significant differences in the grayscale change patterns of k and k'.

[0248] Finally, a probability threshold for difference discrimination is set, empirically set to 0.9 (the second preset threshold). When the probability of secondary confirmation of a suspected area is greater than the threshold, it indicates that it is a true spur area (target spur area).

[0249] Furthermore, for the actual burr region, the image with the greatest difference from the symmetrical region is selected as the local burr image. The virtual emission line region on this image is used as the radius of a rectangle with the length of this region as the center of the rectangle. The resulting rectangular region is then used as the local region of the burr. The location of this rectangular region in the image is selected as the local image of the burr.

[0250] For all other real spur areas, the same method was used to obtain local spur images; thus, the best local images of the spurs were obtained.

[0251] Based on the local image of the burr and its surrounding background region, different local threshold sizes for the burr are obtained.

[0252] It should be noted that the image with local spurs is the region where the relative gray level changes significantly, and there is usually no continuous spur region around it. Therefore, the original corresponding spur image region is used to crop the image from the original image of the spur image, with the image size being twice the side length of the original image.

[0253] Furthermore, since the outer edge of the enlarged local spur image is a normal plane without spurs (background area), and the original inner area is the target area containing spurs (foreground area), the two exhibit significant grayscale distinction due to differences in material and light reflection characteristics. Moreover, the enlarged image precisely forms a bimodal grayscale distribution feature of background (large area normal plane) and foreground (small area spurs + original area). Therefore, the Otsu's Method (OTSU) algorithm can be directly applied to the enlarged local image. The algorithm automatically maximizes the inter-class variance between the foreground and background, thereby determining the optimal spur segmentation threshold grayscale (adaptive grayscale threshold), achieving accurate segmentation without manual intervention.

[0254] Step S303: Perform local image segmentation based on the local image of each burr and its segmentation threshold.

[0255] After obtaining the local spur image and its segmentation threshold, since some spurs exhibit bright characteristics while others exhibit dim characteristics, it is necessary to determine whether the spurs exhibit bright or dim characteristics during segmentation. The local area surrounding the spur is usually a normal planar area. The grayscale of the surrounding local area is diluted by the grayscale of the spur. The closer to the core of the spur area, the lower the degree of grayscale dilution by the spur.

[0256] Therefore, when the gray level changes from a larger local area to a smaller local area, if the gray level of the local area gradually decreases, it indicates that the burr is in low brightness, and pixels with a gray level less than the threshold are burr pixels. Conversely, if the gray level increases when changing from a smaller local area to a larger local area, it indicates that the burr is in high brightness, and pixels with a gray level greater than the segmentation threshold are burr pixels.

[0257] After obtaining the local spiky region image and the threshold size of its segmented image, the average gray value (first average gray value) of the original local region image (first predetermined range) is obtained, and the average gray value (second average gray value) of the local region (second predetermined range) whose respective side lengths are reduced by half is calculated; and the former is denoted as... The latter is For example, the original local area image size is 0-100 pixels * 0-100 pixels, and after being reduced, it becomes a 25-75 pixel * 25-75 pixel area.

[0258] Segmenting of burrs is performed based on a segmentation threshold for the local burr image. Then the segmented white pixels are spiky pixels; if If so, the black pixels after segmentation are spiky pixels.

[0259] This invention proposes a variable illumination condition detection method for precision circular electronic structural components. By changing the illumination conditions, burrs can be identified in detail, thereby obtaining a burr segmentation image under optimal illumination conditions (determined based on grayscale difference metric). This avoids the drawback of poor segmentation effect of a single illumination condition for some burrs, resulting in a higher accuracy rate for burr identification.

[0260] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0261] Reference Figure 9 The diagram illustrates a structural block diagram of a vision-assisted burr detection system for precision electronic components according to the present invention. The vision-assisted burr detection system 60 for precision electronic components specifically includes:

[0262] The acquisition module 601 is used to acquire surface image sequences corresponding to electronic precision components under different lighting conditions and grayscale abrupt change values ​​in the surface image sequences; the grayscale abrupt change values ​​are used to characterize the degree of grayscale abrupt change along adjacent segments of the virtual emission line;

[0263] The determination module 602 is used to determine the suspected burr region based on the grayscale abrupt change value;

[0264] The acquisition unit is also used to acquire the difference in grayscale change between the suspected burr area and its symmetrical area under different lighting conditions;

[0265] The determining unit is further configured to determine the target burr region based on the grayscale change difference value; the symmetrical region and the suspected burr region are symmetrical about the center line of the light source;

[0266] The module 603 is used to perform local image segmentation on the image to be segmented where the target burr region is located based on the brightness and darkness features and the adaptive grayscale threshold, so as to obtain the target burr segmentation image.

[0267] In summary, this invention provides a vision-assisted burr detection system for electronic precision components. During burr detection, the system first acquires surface image sequences and grayscale abrupt change values ​​from these sequences under different lighting conditions. These grayscale abrupt change values ​​characterize the degree of grayscale abrupt change along adjacent segments of a virtual emission line. Based on these values, suspected burr regions are identified. The system then acquires the grayscale variation difference between the suspected burr region and its corresponding symmetrical region under different lighting conditions, and determines the target burr region based on this difference. The symmetrical region and the suspected burr region are symmetrical about the light source centerline. The system determines the brightness and darkness features of the target burr region and its adaptive grayscale threshold. Based on these features and the adaptive grayscale threshold, the system performs local image segmentation on the image to be segmented, obtaining a segmented image of the target burr. This invention allows for a first detection based on grayscale abrupt change values ​​to identify suspected burr regions, followed by a second detection based on grayscale variation differences to determine the authenticity of the suspected burr regions, thereby improving the accuracy of identification.

[0268] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0269] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0270] This invention provides an electronic device for burr detection of precision electronic components under vision assistance. The electronic device includes a memory and one or more programs, which are stored in the memory and configured to be executed by one or more processors. The programs contain instructions for performing the following operations: During burr detection of precision electronic components, firstly, surface image sequences and grayscale abrupt change values ​​in the surface image sequences corresponding to the precision electronic components under different lighting conditions are acquired; the grayscale abrupt change values ​​are used to characterize the degree of grayscale abrupt change along adjacent segments of a virtual emission line; based on the grayscale abrupt change values, suspected burr regions are determined; the grayscale change difference values ​​between the suspected burr regions and the corresponding symmetrical regions under different lighting conditions are acquired, and the target burr region is determined based on the grayscale change difference values; the symmetrical regions and the suspected burr regions are symmetrical about the center line of the light source; the brightness and darkness features of the target burr region and the adaptive grayscale threshold of the target burr region are determined; local image segmentation is performed on the image to be segmented where the target burr region is located based on the brightness and darkness features and the adaptive grayscale threshold to obtain a segmented image of the target burr. In this embodiment of the invention, a first detection can be performed based on the grayscale change value to identify suspected burr areas. Then, a second detection is performed on the suspected burr areas based on the grayscale change difference value to determine whether the suspected burr areas are genuine or not, thereby improving the accuracy of identification.

[0271] This invention also provides an electronic device. Please refer to [link to relevant documentation]. Figure 10 The electronic device may include a processor 501, a memory 502, and a program 5021 stored in the memory 502 and capable of running on the processor 501.

[0272] When program 5021 is executed by processor 501, it can achieve the following: Figure 1 Any steps in the corresponding method embodiments and the achievement of the same beneficial effects will not be repeated here.

[0273] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by hardware related to program instructions, and the program can be stored in a readable medium.

[0274] This invention also provides a readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described functions. Figure 1 Any step in the corresponding method embodiment can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0275] The computer-readable storage medium of this invention can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0276] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0277] The program code contained on the storage medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0278] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or terminal. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0279] This invention also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned steps to achieve the vision-assisted burr detection method for electronic precision components provided in the above embodiments.

[0280] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0281] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0282] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0283] The above provides a detailed description of the artificial intelligence-based fundus lesion image analysis and diagnosis system provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for detecting burrs in precision electronic components with visual assistance, characterized in that, The method includes: Obtain surface image sequences of electronic precision components under different lighting conditions and grayscale abrupt change values ​​in the surface image sequences; the grayscale abrupt change values ​​are used to characterize the degree of grayscale abrupt change along adjacent segments of the virtual emission line; Based on the grayscale abrupt change values, suspected burr areas were identified; The grayscale variation difference between the suspected burr region and the corresponding symmetrical region under different lighting conditions is obtained, and the target burr region is determined based on the grayscale variation difference value; the symmetrical region and the suspected burr region are symmetrical about the center line of the light source. The light and dark features of the target burr region and the adaptive grayscale threshold of the target burr region are determined; the light and dark features are used to characterize the brightness attributes of the target burr region under target illumination conditions; Based on the brightness and darkness features and the adaptive grayscale threshold, local image segmentation is performed on the image to be segmented where the target burr region is located, to obtain the target burr segmentation image.

2. The method for detecting burrs in precision electronic components under vision assistance according to claim 1, characterized in that, The step of determining suspected spurious regions based on the grayscale mutation values ​​includes: For any image in the surface image sequence, each virtual emission line is divided into multiple equal-length sub-segments, and the average gray value of each sub-segment is calculated; Based on the average gray value of each sub-segment, determine the gray value abrupt change between adjacent segments; Based on the grayscale abrupt change value, the probability of suspected burrs is determined; Based on the probability of suspected burrs and the first preset threshold, the suspected burr region is determined.

3. The method for detecting burrs in precision electronic components under vision assistance according to claim 1, characterized in that, The step of obtaining the difference in grayscale variation between the suspected burr region and its symmetrical region under different lighting conditions, and determining the target burr region based on the difference in grayscale variation, includes: For each suspected burr area, determine the illumination symmetry region of the suspected burr area with respect to the center line of the light source; Under different lighting conditions, obtain the average gray value sequence corresponding to the suspected burr region and its lighting symmetrical region. Based on the average gray value sequence, determine the confidence probability of the suspected spurious region; The target burr region is determined based on the confidence probability and the second preset threshold.

4. The method for detecting burrs in precision electronic components under vision assistance according to claim 1, characterized in that, Determining the light and dark features of the target burr region includes: Calculate the first average gray value of a first predetermined range centered on the target burr region; the first predetermined range includes the target burr region and is a closed region whose area is greater than the area of ​​the target burr region. Calculate the second average grayscale value of a second predetermined range that is entirely within the first predetermined range and smaller than the first predetermined range; the second predetermined range includes the target burr region. The light and dark features of the target burr region are determined based on the first average gray value and the second average gray value.

5. The method for detecting burrs in precision electronic components under vision assistance according to claim 4, characterized in that, Determining the light and dark features of the target burr region based on the first average gray value and the second average gray value includes: If the first average gray value is less than the second average gray value, the light and dark features corresponding to the target burr area are determined to be highlight features; If the first average gray value is greater than or equal to the second average gray value, the light and dark features corresponding to the target burr region are determined to be low dark features.

6. The method for detecting burrs in precision electronic components under vision assistance according to claim 1, characterized in that, The adaptive grayscale threshold for determining the target burr region includes: Obtain the image region corresponding to the first predetermined range; The adaptive grayscale threshold is determined based on the grayscale distribution of the image region.

7. The method for detecting burrs in precision electronic components under vision assistance according to claim 1, characterized in that, Before determining the light and dark features of the target burr region and the adaptive grayscale threshold of the target burr region, the method further includes: For each image in the surface image sequence, a grayscale difference metric between the target burr region and the illumination symmetry region in the image is determined; the grayscale difference metric is used to characterize the absolute difference between the average grayscale values ​​of the target burr region and the illumination symmetry region in the same surface image. The image to be segmented is determined from the surface image sequence based on the grayscale difference metric.

8. The method for detecting burrs in precision electronic components under vision assistance according to claim 5, characterized in that, The method further includes: When the target burr region is the highlighted feature, pixels in the image to be segmented whose grayscale value is greater than or equal to the adaptive grayscale threshold are marked as burr pixels. When the target spur region is the low-dark feature, pixels in the image to be segmented whose gray value is less than the adaptive gray value threshold are marked as spur pixels; Based on the burr pixels, the target burr segmentation image is obtained.

9. The method for detecting burrs in precision electronic components under vision assistance according to claim 1, characterized in that, The method further includes: Isolated noise pixels in the target spur segmentation image are removed; the isolated noise pixel is a single pixel whose gray value is inconsistent with the pixel gray value within a preset neighborhood range. The denoised target burr segmentation image is subjected to edge smoothing processing, and the number, area and distribution location information of the target burrs are counted to generate a detection report.

10. A vision-assisted burr detection system for precision electronic components, characterized in that, The system includes: The acquisition module is used to acquire surface image sequences of electronic precision components under different lighting conditions and grayscale abrupt change values ​​in the surface image sequences; the grayscale abrupt change values ​​are used to characterize the degree of grayscale abrupt change along adjacent segments of the virtual emission line; The determination module is used to determine suspected burr regions based on the grayscale abrupt change values; The acquisition unit is also used to acquire the difference in grayscale change between the suspected burr area and its symmetrical area under different lighting conditions; The determining module is further configured to determine the target burr region based on the grayscale change difference value; the symmetrical region and the suspected burr region are symmetrical about the center line of the light source; determine the brightness characteristics of the target burr region and the adaptive grayscale threshold of the target burr region; the brightness characteristics are used to characterize the brightness attributes of the target burr region under the target illumination conditions; The module is used to perform local image segmentation on the image to be segmented where the target burr region is located based on the brightness and darkness features and the adaptive grayscale threshold, so as to obtain the target burr segmentation image.

Images