A method for surface quality testing of steel strands used in mining anchor cables after stabilization treatment.

Abstract

This invention relates to the field of image data processing technology, and more specifically, to a method for surface quality inspection of mining anchor cable steel strands after stabilization treatment. The method includes: acquiring a surface image of the steel strand to be inspected; determining a local texture coherence index for each pixel; determining a longitudinal consistency deviation index for each pixel; determining a defect saliency index for each pixel; and identifying defect regions in the surface image of the steel strand to be inspected based on the defect saliency index of each pixel. This invention constructs local texture coherence and longitudinal consistency deviation indices, combining grayscale variation characteristics and spatial variation rate to achieve multi-dimensional quantitative detection of minute defects on the surface of the steel strand. This effectively distinguishes between noise and real defects, improves sensitivity to low-contrast defects, reduces missed detections and false alarms, and provides a highly reliable quality inspection method for mining anchor cable steel strands.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology. More specifically, this invention relates to a method for detecting the surface quality of mining anchor cable strands after stabilization treatment. Background Technology

[0002] As a core load-bearing component in the support system of underground engineering projects such as coal mine roadways, the manufacturing quality of mining anchor cable steel strand directly determines the long-term stability and safety of the project. Stabilization treatment is a crucial step in the steel strand production process. This treatment eliminates internal stress in the material through high-temperature stretching and shaping, significantly improving its mechanical properties. However, this high-temperature, high-stress process inevitably introduces various minute defects into the surface of the steel strand, such as tiny oxide spots, shallow scratches, and fine pitting caused by uneven local stress. These defects are small in size and have extremely low contrast with the background of the steel strand, but they are potential stress concentration points and corrosion initiation points, posing serious safety hazards to the long-term service performance and fatigue life of the anchor cable.

[0003] Currently, the industry widely adopts automated inspection technology based on machine vision to replace manual visual inspection, thereby improving inspection efficiency and consistency. This type of technology typically uses industrial cameras to acquire images of the steel strand surface, and then utilizes image processing algorithms or deep learning models to identify defects.

[0004] However, steel strands are composed of multiple strands of steel wire spirally twisted together, and their surface naturally exhibits complex and regular periodic textures. The visual features of subtle defects (such as grayscale variations and shape characteristics) are often very similar to the texture features of the steel strands themselves, making them easily obscured by strong background textures. This results in a high false negative rate for traditional algorithms based on fixed thresholds, edge detection, or morphological processing. Furthermore, while deep learning-based detection methods perform well on specific defect types, their ability to identify small, low-contrast defects remains limited. Deep learning models are prone to overfitting to normal textures, misclassifying normal texture fluctuations as defects, leading to a high false positive rate. Moreover, because subtle defects are low-probability events in actual production, defect samples are scarce, making it difficult for models to fully learn their effective features, thus leading to false negatives and affecting the reliability of quality inspection of mining anchor cable steel strands. Summary of the Invention

[0005] To address the technical problem of high missed detection rate and high false alarm rate of minute defects caused by strong texture interference from the steel strand itself, this invention proposes a surface quality inspection method for mining anchor cable steel strands after stabilization treatment. This method includes the following steps:

[0006] Acquire a surface image of the steel strand to be inspected; designate any pixel in the surface image as a target pixel; determine a local texture coherence index for the target pixel to characterize the degree of consistency of local texture direction based on the texture direction of each pixel in the preset neighborhood of the target pixel; determine a longitudinal consistency deviation index for the target pixel to characterize the degree of local gray-scale abrupt change based on the gray-scale distribution difference of multiple pixel blocks along the texture direction of the target pixel and the local texture coherence index; determine a defect saliency index for the target pixel based on the longitudinal consistency deviation index and the spatial change rate of the longitudinal consistency deviation index; and identify defect regions in the surface image of the steel strand to be inspected based on the defect saliency index of each pixel.

[0007] This invention constructs a local texture coherence index by analyzing the texture direction of each pixel in the neighborhood of a target pixel. This index effectively quantifies the directional consistency characteristics of the steel strand surface texture, thereby distinguishing normal texture areas from abnormal areas and providing a reliable background reference for subsequent defect detection. By analyzing the grayscale distribution differences of multiple pixel blocks along the texture direction and combining it with the local texture coherence index, a longitudinal consistency deviation index is constructed. This index effectively captures grayscale abrupt changes along the texture direction, avoiding potential misjudgments from traditional omnidirectional detection and improving sensitivity to subtle defects. By comprehensively considering the longitudinal consistency deviation and its spatial variation rate through a defect saliency index, a multi-dimensional quantitative evaluation of defect features is achieved. The introduction of the spatial variation rate effectively distinguishes local noise from real defects, avoiding false alarms that may occur with a single index. By fully utilizing the regularity characteristics of the steel strand surface texture and organically combining texture coherence and grayscale consistency deviation, the detection capability for small, low-contrast defects is improved. This effectively solves the problems of high false negative rates in traditional methods and high false alarm rates in deep learning methods, providing a more reliable technical guarantee for the quality inspection of mining anchor cable steel strands.

[0008] Furthermore, acquiring the surface image of the steel strand to be tested includes: acquiring images from multiple perspectives distributed around the axis of the steel strand using an image acquisition device; stitching the images from multiple perspectives together and performing coordinate transformation to unfold the cylindrical surface image of the steel strand into a two-dimensional planar image, thereby obtaining the surface image of the steel strand to be tested.

[0009] Furthermore, the method for obtaining the local texture coherence index is as follows: construct a neighborhood window centered on the target pixel; calculate the local texture direction angle of each pixel within the neighborhood window; perform frequency doubling on the local texture direction angle and map it into a complex unit vector; perform vector summation and averaging on all complex unit vectors within the neighborhood window, and use the magnitude of the averaged vector as the local texture coherence index of the target pixel.

[0010] This invention effectively captures the local directional features of the surface texture of steel strand by constructing a neighborhood window and calculating the local texture direction angle of each pixel. The frequency doubling process maps the texture direction angle to the complex domain, making full use of the periodicity of the texture and enhancing the sensitivity of the direction consistency detection. By mapping the texture direction angle after frequency doubling to a complex unit vector and performing vector summation and averaging, a quantitative index of local texture coherence is constructed, which can effectively distinguish between normal areas with consistent texture direction and abnormal areas with chaotic texture direction. The closer the vector magnitude is to 1, the higher the texture coherence; the closer it is to 0, the more chaotic the texture direction.

[0011] Furthermore, the method for obtaining the local texture direction angle is as follows: for each pixel in the neighborhood window, construct the Hessian matrix of the pixel; perform eigenvalue decomposition on the Hessian matrix of the pixel to obtain the eigenvector corresponding to the minimum eigenvalue; take the direction of the eigenvector corresponding to the minimum eigenvalue as the local texture direction of the pixel, calculate the angle between the local texture direction and the preset coordinate axis, and obtain the texture direction angle of the pixel.

[0012] Furthermore, the longitudinal consistency deviation index is obtained as follows: a rectangular pixel block is constructed with the target pixel as the center pixel block, and multiple path pixel blocks of the same size as the center pixel block are constructed on both sides of the center pixel block along the local texture direction of the target pixel; the desired gray-level histogram is calculated based on the gray-level histograms of the multiple path pixel blocks; the dispersion difference moment of the target pixel is calculated based on the difference between the gray-level histogram of the center pixel block and the desired gray-level histogram; and the longitudinal consistency deviation index of the target pixel is determined by combining the dispersion difference moment of the target pixel with the local texture coherence index.

[0013] This invention achieves precise tracking of texture direction by constructing a central pixel block and multiple path pixel blocks along the local texture direction. It can effectively capture grayscale change features along the texture direction and avoid interference that may occur in traditional omnidirectional analysis. By calculating the expected grayscale histogram of multiple path pixel blocks as a reference benchmark, it can accurately reflect the grayscale distribution pattern of normal texture areas. The difference calculation between the central pixel block and the expected histogram (dispersion difference moment) can effectively identify local grayscale anomalies and improve the detection sensitivity of subtle defects. By combining the dispersion difference moment and the local texture coherence index to determine the longitudinal consistency deviation index, it achieves an organic combination of texture background and grayscale anomalies. When the texture coherence is high, it indicates that the texture of the area is highly regular, and any grayscale anomaly is more likely to be a real defect. When the texture coherence is low, it indicates that the texture of the area itself is chaotic, and a higher grayscale anomaly threshold is needed to avoid false alarms.

[0014] Further, the calculation of the desired grayscale histogram includes: taking a Gaussian weighted average of the actual grayscale histograms of multiple path pixel blocks based on the distance between each path pixel block and the center pixel block to obtain the desired grayscale histogram.

[0015] Furthermore, the longitudinal consistency deviation index satisfies:

[0016] In the formula, For pixels The longitudinal consistency deviation index. For pixels The dispersion difference moment, For pixels The local texture coherence index, It is the hyperbolic tangent function.

[0017] This invention achieves saturation processing of the longitudinal consistency deviation index by nonlinearly mapping the dispersion difference moment using the hyperbolic tangent function, avoiding interference from extreme values ​​in subsequent processing, while maintaining sensitivity to moderate anomalies. By organically combining the dispersion difference moment with the texture coherence, it improves the ability to identify minute defects and resist noise interference, effectively solving the problem of both missed detections and false alarms in the surface quality inspection of steel strands.

[0018] Furthermore, the dispersion difference moment satisfies:

[0019] In the formula, For pixels The dispersion difference moment, Maximum gray level For grayscale indexes, It is the average gray level of the expected gray level histogram of the center pixel block. It is the actual histogram of the center pixel block in grayscale. The value, It is the expected gray-level histogram of the center pixel block at the gray level. The value, It is the absolute value symbol.

[0020] This invention constructs a quantitative model of dispersion difference moment by calculating the absolute value of the difference between the actual histogram and the expected histogram of the center pixel block at each gray level, and by weighting the difference between the gray level and the average gray value. This model can effectively measure the degree of deviation of the gray distribution of the center pixel block from the normal texture area. The weighting factor reflects the influence of the degree of gray level deviation from the center. The greater the gray level difference that deviates from the average gray level, the greater its contribution to the dispersion difference moment. This is consistent with the physical law that defect characteristics are usually manifested as extreme gray value changes. The absolute value calculation ensures that both positive and negative differences can effectively contribute to the difference measurement.

[0021] Furthermore, the defect significance index satisfies:

[0022] In the formula, For pixels The defect significance index For pixels The longitudinal consistency deviation index. for gradient vector, for The length of the module.

[0023] This invention uses the product of the longitudinal consistency deviation index and its gradient magnitude as a defect significance index, thus achieving an organic combination of local anomaly intensity and spatial variation rate. The longitudinal consistency deviation index reflects the degree to which a pixel deviates from the normal texture, while the gradient magnitude characterizes the severity of the spatial variation of the anomaly. The product form ensures that only regions with both high-intensity anomalies and significant spatial variations will obtain high significance scores. A high deviation value alone may be caused by noise, while a high gradient value alone may be generated by the boundary of the normal texture. By comprehensively considering the anomaly intensity and spatial variation characteristics, the accuracy and robustness of defect detection are improved, providing reliable support for the detection of surface quality of steel strands for mining anchor cables.

[0024] Furthermore, the step of identifying defect regions in the surface image of the steel strand to be inspected based on the defect saliency index of each pixel includes: generating a saliency map composed of the defect saliency index of each pixel; performing adaptive threshold segmentation on the saliency map using the maximum inter-class variance method to obtain a binarized image; and determining the regions in the binarized image whose pixel values ​​are identified as defects as defect regions in the surface image of the steel strand to be inspected, thus completing the surface quality inspection method for the stabilized steel strand of mining anchor cables.

[0025] The present invention has the following beneficial effects:

[0026] (1) Breaking through the limitations of traditional algorithms in distinguishing between subtle defects and the spiral texture of the steel strand itself, the consistency of texture direction is quantified by the local texture coordination index. Normal texture areas exhibit highly consistent directional characteristics due to the spiral structure, while defect areas will destroy this coordination. Combined with the longitudinal consistency deviation index to capture gray-scale abrupt changes along the texture direction, subtle defects (such as small scratches and pits) can be effectively separated from the periodic texture background, solving the problem of missed detection caused by defects being submerged by texture, and improving the recognition sensitivity of low-contrast and small-sized defects.

[0027] (2) By integrating the longitudinal consistency deviation and its spatial variation rate through the defect significance index, the feature that “real defects have local gray-scale abrupt changes and continuous spatial distribution” is strengthened, while the periodic gray-scale fluctuation of normal texture is weakened. Compared with the problem that deep learning models are prone to misjudging texture fluctuations as defects, it can accurately filter out false defect signals caused by uneven texture and reduce the false alarm rate. It is especially suitable for scenarios where the surface texture of steel strand is complex but the defect features are weak.

[0028] (3) It does not require a large number of labeled defect samples for model training. Instead, it constructs detection indicators based on the inherent characteristics of steel strand texture (directional consistency, gray scale distribution pattern), which solves the problem of insufficient model learning caused by the scarcity of minor defect samples in actual production. It has low requirements for the number of samples and can be directly applied to real-time detection on the production line, adapting to the industrial scenario requirements of quality inspection of mining steel strand.

[0029] (4) By accurately identifying minute surface defects, potential quality hazards (such as stress concentration points and material damage) can be discovered in a timely manner after the steel strand stabilization treatment, avoiding the use of unqualified products in mine anchor cable projects. This not only improves the quality control accuracy of steel strand production, but also reduces the risks of mine support failure and safety accidents caused by material defects. At the same time, it reduces rework costs caused by missed inspections and misjudgments, and improves production efficiency. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating the steps of a method for detecting the surface quality of mining anchor cable steel strand after stabilization treatment, according to an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below. The described embodiments are only a part 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.

[0032] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] Please see Figure 1The diagram illustrates a flowchart of a method for detecting the surface quality of mining anchor cable steel strand after stabilization treatment, according to an embodiment of the present invention. The method includes the following steps:

[0034] S1: Obtain the surface image of the steel strand to be inspected.

[0035] It should be noted that this step aims to acquire two-dimensional image data that comprehensively reflects the surface condition of the steel strand and is easy to process subsequently. Since the steel strand is a cylindrical structure, a single camera capturing images from a single angle cannot obtain complete surface information. Therefore, in this embodiment, multiple industrial cameras arranged in a ring or a line array camera combined with the rotation of the steel strand can be used to acquire a ring-shaped field-of-view image array centered on the steel strand's axis, thereby obtaining circumferential image data of the steel strand's surface.

[0036] Specifically, acquiring the surface image of the steel strand to be inspected includes:

[0037] Images from multiple perspectives distributed around the axis of the steel strand are acquired using image acquisition equipment;

[0038] Images from multiple perspectives are stitched together and subjected to coordinate transformation (to eliminate perspective distortion and uneven lighting caused by the cylindrical curved surface), and the surface image of the cylindrical steel strand is unfolded into a two-dimensional planar image to obtain the surface image of the steel strand to be tested.

[0039] S2: Determine the local texture coherence index for each pixel.

[0040] It should be noted that this step is based on prior knowledge of the physical structure of steel strands. In defect-free areas, the surface texture of steel strands is mainly composed of regularly arranged steel wires, so its local texture direction should exhibit a high degree of coordination. The presence of defects will disrupt this regular arrangement, leading to disorder in the texture direction. The purpose of this step is to quantify this physical "coordination" or "disorder" into a specific index.

[0041] Any pixel in the surface image is designated as the target pixel. Based on the texture direction of each pixel in the preset neighborhood of the target pixel, a local texture coherence index is determined to characterize the degree of consistency of the local texture direction of the target pixel.

[0042] Specifically, firstly, with the target pixel point Built around rectangular neighborhood window (For example, ).

[0043] Next, for the neighborhood window Any pixel within The local structure of an image is analyzed by calculating its Hessian matrix. The Hessian matrix is ​​a square matrix composed of second derivatives (obtained through the finite difference method for each pixel), reflecting the curvature changes of image grayscale locally. The eigenvectors of this matrix indicate the directions of fastest and slowest grayscale changes. For linear or edge-like structures, grayscale changes slowest along their direction and fastest perpendicular to it. Therefore, the eigenvector corresponding to the smaller eigenvalue of the Hessian matrix can be considered as the eigenvector of that pixel. The local texture principal direction vector, the angle between this vector and the horizontal axis is denoted as . .

[0044] Then, considering the direction and Physically representing the same linear texture, to eliminate this mathematical ambiguity, this embodiment employs an angle frequency multiplication and mapping method to the complex plane; each direction angle... Convert to a complex unit vector, i.e. Through this transformation, and The direction angles are all mapped onto the complex plane. The location solves the problem of ambiguity in direction.

[0045] Finally, calculate the window. The average of the complex unit vectors corresponding to all pixels within the range, with its magnitude taken as the value of the pixel. Local texture coherence index , satisfy:

[0046] ;

[0047] In the formula, For neighborhood windows The number of pixels within, It is an imaginary number. This is the modulus symbol.

[0048] In the intact wire texture area, the orientation angle of all pixels. Because the texture orientations of all pixels within a neighborhood window are highly consistent, their corresponding complex vectors will point in the same direction. When summing these vectors, they will accumulate along the same direction, resulting in a very long resultant vector. Even after averaging, the endpoint of the average vector will still be very close to the boundary of the unit circle. Therefore, the magnitude of its average vector (i.e., ...) is very long. The result will approach 1; conversely, if there are defects within the window that cause the texture orientation to be disordered, the complex vectors will point in all directions. When these vectors are added together, they will cancel each other out. For example, a rightward vector will be canceled out by a leftward vector, and an upward vector will be canceled out by a downward vector, making the result... Approaching 0.

[0049] S3: Determine the longitudinal consistency deviation index for each pixel.

[0050] It should be noted that this step aims to utilize another key physical prior: along the axial direction of a single steel wire (i.e., the dominant structural direction), its surface material and light reflection characteristics should maintain a high degree of continuity and consistency in the absence of defects. Tiny oxide spots or scratches will disrupt this consistency. This step effectively amplifies local abrupt changes caused by defects by comparing the feature differences of a pixel with its neighboring points before and after the wire's direction.

[0051] Based on the differences in grayscale distribution of multiple pixel blocks along the texture direction of the target pixel and the local texture coherence index, a longitudinal consistency deviation index for characterizing the degree of local grayscale abrupt change in the target pixel is determined.

[0052] Specifically, firstly, for the target pixel... Build a with it as the center The center pixel block (e.g.) Simultaneously, based on the local texture direction angle determined in the above steps... Samples were taken before and after along this direction. Two of the same size The path pixel block.

[0053] Next, a "desired" grayscale distribution model needs to be constructed for the central pixel block. This model is achieved by... The expected gray-level histogram is obtained by weighting the normalized gray-level histograms of each path pixel block. , satisfy:

[0054] ;

[0055] In the formula, This represents the number of the path pixel block. For the first The weights of the grayscale histogram of each path pixel block For the first Gray-level histogram of each path pixel block.

[0056] Among them, weight The design ensures that path pixels closer to the center pixel block contribute more. For example, Gaussian weighting can be used: , For the first From the center point of each path pixel block to the target pixel point The distance is obtained in this way. This represents the "normal" grayscale distribution in the absence of defects, inferred from the context.

[0057] Then, calculate the grayscale histogram of the center pixel block. With expected grayscale histogram The difference between them is addressed in this embodiment using the dispersion difference moment. To measure, satisfy:

[0058] ;

[0059] In the formula, For pixels The dispersion difference moment, Maximum gray level For grayscale indexes, It is the average gray level of the expected gray level histogram of the center pixel block. It is the actual histogram of the center pixel block in grayscale. The value, It is the expected gray-level histogram of the center pixel block at the gray level. The value, It is the absolute value symbol.

[0060] Among them, for This part calculates the pixel frequency of the actual observed gray level at each gray level (i.e., ) and the pixel frequency of the desired grayscale (i.e. It measures the absolute magnitude of the difference, only quantifying the "amplitude of the difference," without considering whether the difference is increasing or decreasing. For This is a dynamically calculated weight, which physically assigns an "importance" factor to the difference between each gray level. This importance is proportional to the distance of that gray level from the "average gray level of the desired gray level". Distance from average gray level The farther away, the larger the weight value, and vice versa; this design gives high importance to differences occurring in extremely bright or dark areas, while differences occurring near the average brightness are considered secondary information. For the formula as a whole, if the center pixel block is in a normal, defect-free area, due to the longitudinal consistency of the steel strand grayscale, and Highly similar Very small, calculated The histogram tends to 0; if the center pixel block is a region with minor defects and the path pixel blocks are defect-free regions, the histogram... This will be compared with the expected grayscale histogram. Significant differences occurred. The differences will be significant, and these differences will be... Magnify it further to output a very large value. The value. It should be added that if both the center pixel block and the path pixel block are areas with minor defects, the calculated value... It will be very small, but this is not a concern at this point, because the purpose of this invention is to highlight "minor defects." When the pixel blocks are all defective areas, they are already quite noticeable in the image, and no further enhancement is needed; furthermore, although It describes the characteristics of the center pixel block, but the center pixel block is based on the target pixel point. Constructed, therefore Can be used as target pixel The characteristics, when When the size increases, the central pixel block is considered a defect area, because As the center point of the central pixel block, the target pixel point The likelihood of it being a defect also increases, therefore, Used to reflect target pixels The possibility of it being a defect.

[0061] Finally, the dispersion difference moment Coordination index with local texture By merging the data, we can obtain the longitudinal consistency deviation index. , satisfy:

[0062] ;

[0063] In the formula, For pixels The longitudinal consistency deviation index. It is the hyperbolic tangent function.

[0064] in, Used to Normalization, range restricted to , It acts as a weight for "structural credibility". For the entire formula, a point is considered a significant deviation (i.e., A high gray value (close to 1) requires two conditions to be met simultaneously: its gray value has a significant relative difference from the surrounding area; and the structure of the local area in which it is located is chaotic and uncoordinated. This can greatly improve the reliability of the indicator and effectively avoid misjudging normal texture fluctuations as deviations.

[0065] S4: Determine the defect significance index for each pixel.

[0066] It should be noted that the purpose of this step is to further refine the defect signal and suppress large-scale, slowly varying artifacts that may be caused by factors such as uneven illumination. The core logic is that a real, subtle defect is not only an anomaly in itself, but its appearance is also "abrupt," meaning that there is a steep boundary between the abnormal region and the normal region.

[0067] Based on the longitudinal consistency deviation index and the spatial variation rate of the longitudinal consistency deviation index, the defect significance index of the target pixel is determined.

[0068] Specifically, the defect significance index satisfies:

[0069] ;

[0070] In the formula, For pixels The defect significance index For pixels The longitudinal consistency deviation index. for The gradient vector (which can be approximated by operators such as Sobel). for The length of the module.

[0071] in, Represents pixels The degree of deviation, or the "initial probability" of a defect, is not perfect; when a real, minute defect exists in an area, The pixels on the defect will have a higher value, and at the edge of the defect, The value will abruptly change from high to low, causing its gradient magnitude to... Similarly, the product of the two will make the defective pixel... The pixels are significantly magnified; however, for pixels in areas with large, gradually changing deviations caused by uneven lighting, although... It may have a certain value, but its gradient magnitude is very small, and the result after multiplication is... The value will be suppressed. Therefore, only those that are both abnormal ( The value is relatively large and then suddenly ( Only tiny defective pixels (with larger values) will output a larger value. This further amplifies the "isolated and abrupt" defect characteristics of real micro-defects.

[0072] S5: Based on the defect significance index of each pixel, identify the defect area in the surface image of the steel strand to be inspected.

[0073] Specifically, identifying defect regions in the surface image of the steel strand to be inspected based on the defect saliency index of each pixel includes:

[0074] Generate a saliency map composed of the defect saliency index of each pixel;

[0075] By using an adaptive thresholding method, such as the maximum inter-class variance method, to binarize and segment the saliency map, the mask of all defect regions can be accurately segmented.

[0076] By transforming the segmented mask coordinates back to the original, unexpanded image coordinate system through inverse coordinate transformation, all identified minor defects can be located and marked on the original image, thus completing the surface quality inspection method after the stabilization treatment of mining anchor cable steel strands.

[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting the surface quality of mining anchor cable steel strand after stabilization treatment, characterized in that, include: Acquire a surface image of the steel strand to be inspected; Any pixel in the surface image is designated as the target pixel. Based on the texture direction of each pixel in the preset neighborhood of the target pixel, the local texture coherence index of the target pixel is determined to characterize the degree of consistency of local texture direction. Based on the differences in grayscale distribution among multiple pixel blocks along the texture direction of the target pixel and the local texture synergy index, a longitudinal consistency deviation index for characterizing the degree of local grayscale abrupt change in the target pixel is determined. The longitudinal consistency deviation index is obtained as follows: a rectangular pixel block is constructed with the target pixel as the center pixel block, and multiple path pixel blocks of the same size as the center pixel block are constructed on both sides of the center pixel block along the local texture direction of the target pixel. The desired grayscale histogram is calculated based on the grayscale histograms of the multiple path pixel blocks. The dispersion difference moment of the target pixel is calculated based on the difference between the grayscale histogram of the center pixel block and the desired grayscale histogram. The longitudinal consistency deviation index of the target pixel is determined by combining the dispersion difference moment of the target pixel with the local texture synergy index. The longitudinal consistency deviation index satisfies: ； For pixels The longitudinal consistency deviation index. For pixels The dispersion difference moment, For pixels The local texture coherence index, The hyperbolic tangent function is used; the dispersion difference moment is used to reflect the target pixel. The possibility of it being a defect; The dispersion difference moment satisfies: ； Maximum gray level For grayscale indexes, It is the average gray level of the expected gray level histogram of the center pixel block. It is the actual histogram of the center pixel block in grayscale. The value, It is the expected gray-level histogram of the center pixel block at the gray level. The value, The absolute value sign is used; the defect significance index of the target pixel is determined based on the longitudinal consistency deviation index and the spatial change rate of the longitudinal consistency deviation index. Based on the defect saliency index of each pixel, defect areas in the surface image of the steel strand to be inspected are identified.

2. The method for surface quality testing of mining anchor cable steel strand after stabilization treatment according to claim 1, characterized in that, The process of acquiring the surface image of the steel strand to be inspected includes: Images from multiple perspectives distributed around the axis of the steel strand are acquired using image acquisition equipment; Images from multiple perspectives are stitched together and coordinate transformed to unfold the surface image of the cylindrical steel strand into a two-dimensional planar image, thus obtaining the surface image of the steel strand to be inspected.

3. The method for surface quality testing of mining anchor cable steel strand after stabilization treatment according to claim 1, characterized in that, The method for obtaining the local texture synergy index is as follows: Construct a neighborhood window centered on the target pixel; Calculate the local texture orientation angle of each pixel within the neighborhood window; The local texture direction angle is frequency-doubled and mapped to a complex unit vector; The vector summation and averaging of all complex unit vectors within the neighborhood window are used as the magnitude of the averaged vector as the local texture coherence index of the target pixel.

4. The method for surface quality testing of mining anchor cable steel strand after stabilization treatment according to claim 3, characterized in that, The method for obtaining the local texture direction angle is as follows: For each pixel within the neighborhood window, construct the Hessian matrix for that pixel; Perform eigenvalue decomposition on the Hessian matrix of the pixel to obtain the eigenvector corresponding to the smallest eigenvalue. The direction of the feature vector corresponding to the minimum feature value is taken as the local texture direction of the pixel. The angle between the local texture direction and the preset coordinate axis is calculated to obtain the texture direction angle of the pixel.

5. The method for surface quality testing of mining anchor cable steel strand after stabilization treatment according to claim 1, characterized in that, The calculation to obtain the desired grayscale histogram includes: The desired grayscale histogram is obtained by performing a Gaussian weighted average based on the distance between each path pixel block and the center pixel block on the actual grayscale histograms of multiple path pixel blocks.

6. The method for surface quality testing of mining anchor cable steel strand after stabilization treatment according to claim 1, characterized in that, The defect significance index satisfies: ； In the formula, For pixels The defect significance index For pixels The longitudinal consistency deviation index. for gradient vector, for The length of the module.

7. The method for surface quality testing of mining anchor cable steel strand after stabilization treatment according to claim 1, characterized in that, The method of identifying defect regions in the surface image of the steel strand to be inspected based on the defect saliency index of each pixel includes: Generate a saliency map composed of the defect saliency index of each pixel; The saliency map is adaptively thresholded using the maximum inter-class variance method to obtain a binarized image; In the binarized image, the area marked as a defect by pixel value is identified as the defect area in the surface image of the steel strand to be inspected, thus completing the surface quality inspection method for the stabilized steel strand of mining anchor cable.

Images