Steel wire drawing surface defect detection method and detection equipment based on image fusion

Abstract

The invention relates to the technical field of steel wire detection, and discloses a steel wire drawing surface defect detection method and detection equipment based on image fusion, the detection equipment comprises an equipment base and a detection unit arranged along a conveying path of a to-be-detected steel wire, and an external traction module is arranged along the conveying path of the to-be-detected steel wire; according to the device, the collaborative innovative design of double optical detection modules and reflecting prisms is adopted, the first optical detection module and the second optical detection module are distributed in a staggered mode along the axis of the steel wire to be detected, and the light path steering function of the first reflecting prism and the light path steering function of the second reflecting prism are matched; the defect detection of the full circumferential surface of the steel wire can be realized by adopting the structure, the defects that the traditional multi-module distributed arrangement needs a plurality of high-specification visual assemblies, the optical matching is complicated and the debugging is tedious are abandoned, and meanwhile, due to the light path correction characteristic reflected by the prism, the problem of picture distortion easily occurring when a cylindrical surface is shot by a single camera is effectively reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of steel wire detection, in particular to a steel wire wire-drawing surface defect detection method based on image fusion and a detection device. BACKGROUND

[0002] Steel wire is a core raw material in the fields of mechanical manufacturing, construction engineering, electronic devices, etc., and its surface defects (such as scratches, pits, oxide scales, local uneven thickness, etc.) will directly affect the mechanical properties, corrosion resistance and assembly precision of products. Therefore, high-precision defect detection is required on the surface of the steel wire after the wire-drawing process. The current mainstream detection scheme in the industry is to use machine vision technology to capture images of the steel wire in the conveying process through a vision module, and then to analyze and identify surface defects in combination with an algorithm. This technology is widely used in metal wire quality control due to its high detection accuracy and efficiency.

[0003] However, the existing steel wire surface visual detection device still has the following defects in actual implementation: 1. High cost of circumferential surface detection, difficult installation, debugging and maintenance. In order to achieve full coverage of defects on the 360° circumferential surface of the steel wire, the existing technology often adopts a structure in which multiple vision modules are distributed along the circumferential direction of the steel wire. Although this structure can meet the detection range requirements, it requires multiple high-specification vision modules and complex optical supporting components, significantly increasing the hardware cost. Meanwhile, the installation of multiple modules requires accurate calibration of the spatial angle and relative position of each vision module, which is a tedious debugging process. When the equipment fails later, the relative positions of all modules need to be recalibrated for maintenance, greatly increasing the difficulty and man-hours of maintenance.

[0004] 2. Limited field of view of a single vision module, high defect omission rate. If a single vision module is used for detection, due to the limitations of the planar imaging field of view of the camera and the cylindrical structure of the steel wire, only the local circumferential area (usually less than 180°) of the steel wire can be imaged. The circumferential surface defects on the side facing away from the camera cannot be effectively captured, resulting in an increased omission rate of subtle defects such as scratches and pits, which seriously affects the comprehensive judgment of the surface quality of the steel wire, and may cause the defective steel wire to flow into the downstream process, leading to product quality accidents.

[0005] Therefore, the present application proposes a steel wire wire-drawing surface defect detection method based on image fusion and a detection device. SUMMARY

[0006] The present application aims to provide a steel wire wire-drawing surface defect detection method based on image fusion and a detection device to solve the problems raised in the background.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an image fusion-based steel wire drawing surface defect detection device, comprising a device base and a detection unit arranged along the conveying path of the steel wire to be detected, an external traction module set along the conveying path of the steel wire to be detected, used to provide power for the continuous conveying of the steel wire to be detected, the detection unit comprising a first optical detection module and a second optical detection module, the first optical detection module and the second optical detection module being staggered along the axial direction of the steel wire to be detected, the first optical detection module being used for visual detection of half a circumferential area of ​​the steel wire to be detected, and the second optical detection module being used for visual detection of the other half circumferential area of ​​the steel wire to be detected.

[0008] Preferably, the first optical detection module and the second optical detection module are both mounted on the equipment base and are arranged in a staggered manner along the steel wire conveying path of the equipment base. The first optical detection module includes a first vision module and a first reflecting prism. The first reflecting prism directs the surface reflected light of the steel wire portion that the first vision module cannot directly observe to the first vision module, enabling the first vision module to acquire the reflected image. The second optical detection module includes a second vision module and a second reflecting prism. The second reflecting prism directs the surface reflected light from other parts of the wire that cannot be directly observed by the second vision module to the second vision module, enabling the second vision module to acquire the reflected image and complement the detection area of ​​the first vision module.

[0009] Preferably, both the first and second reflecting prisms are right-angle triangular prisms, and the reflecting surfaces are coated with an anti-reflection film.

[0010] Preferably, the first vision module has two vision ends: one for directly capturing the observable area of ​​the steel wire to be detected, and the other for collecting the blind area of ​​the steel wire after being reflected by the first reflecting prism. The second vision module is equipped with two vision ends: one for directly capturing the image of another observable area of ​​the steel wire to be detected, and the other for collecting the image of the blind area of ​​the steel wire after being reflected by the second reflecting prism.

[0011] Preferably, the device base surface is equipped with a mounting bracket, and both the first and second reflecting prisms are fixedly connected to the mounting bracket.

[0012] Preferably, the device base is equipped with a crossbeam, which is located on the side away from the first optical detection module and the second optical detection module. The steel wire to be tested passes through the crossbeam, and an annular cleaning nozzle is provided on the crossbeam, and the annular cleaning nozzle is coaxially arranged with the steel wire to be tested.

[0013] Preferably, the annular cleaning nozzle is externally connected to an airflow delivery pipe, the other end of which is connected to an external gas source to provide cleaning airflow to the annular cleaning nozzle.

[0014] The method for detecting surface defects in steel wire drawing based on image fusion includes the following steps: S1: Start the external traction module, set the conveying speed threshold of the steel wire to be detected, and simultaneously calibrate the acquisition parameters of the first optical detection module and the second optical detection module—adjust the misalignment distance between the two detection modules according to the diameter of the steel wire to be detected, and calibrate the image acquisition frequency. S2: The external traction module drives the steel wire to be tested to move at a constant speed along the conveying path, while simultaneously triggering the first optical detection module and the second optical detection module to work synchronously. The first optical detection module acquires the direct observation image of the first semi-circular area of ​​the steel wire to be detected and the image completed by prism reflection; The second optical detection module acquires direct observation images and prism reflection-completed images of the second semi-circular region of the steel wire to be detected. The acquisition process of the two detection modules is linked by a synchronous controller on the equipment base to ensure that the images acquired at the same time correspond to the same axial position of the steel wire to be detected; S3: Process the image acquired in S2: Use an adaptive median filtering algorithm to specifically eliminate image noise caused by airflow disturbances and slight vibrations of the steel wire, while preserving defect details; Illumination compensation: By stretching grayscale, local uneven illumination caused by differences in the light path during prism reflection is compensated, so that the grayscale distribution of the image is uniform. Defect Enhancement: Apply edge enhancement algorithms to highlight the edge contours of minute defects on the steel wire surface, thereby improving the sensitivity of subsequent defect identification; Coordinate correction: Based on the preset misalignment distance between the two detection modules, the coordinate mapping algorithm is used to eliminate the axial offset of the image caused by the misalignment distribution along the axis, ensuring that the axial coordinates of the two types of images are accurately aligned. S4: Based on the cylindrical geometric features of the steel wire to be detected, establish an image coordinate mapping model: extract the edge features of the directly observed area in the image of the first optical detection module and the texture features of the reflection completion area in the image of the second optical detection module; By combining weighted average fusion and gradient domain fusion, the preprocessed images of the detection unit are stitched together to form a fully unfolded image of the steel wire to be detected, ensuring that there are no obvious grayscale breaks at the stitching points. S5: The model first locates the suspected defect area, and then extracts the geometric and texture features of the defect; Suspected defects are classified using a pre-defined defect classifier, and the defect type and confidence level are output. S6: Record the specific location information of each defect and generate an inspection report; at the same time, feed the defect data back to the control system of the wire drawing equipment to adjust the drawing die parameters or lubrication dosage to reduce the occurrence of subsequent defects.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. In view of the shortcomings of the prior art, such as high cost, complicated installation and debugging, and easy image distortion of multi-vision module circumferential detection, the present device adopts a collaborative innovative design of dual optical detection modules + reflective prism. By simply distributing the first optical detection module and the second optical detection module along the axis of the steel wire to be detected in a staggered manner, and with the light path turning function of the first reflective prism and the second reflective prism, defect detection of the entire circumference of the steel wire can be realized. This eliminates the disadvantages of the traditional multi-module distributed arrangement that requires multiple high-specification vision components, complex optical matching, and complicated debugging. At the same time, due to the light path correction characteristics of prism reflection, the image distortion problem that is easy to occur when shooting the cylindrical surface with a single camera is effectively reduced, making the edges of the acquired image clearer and the geometric features more accurate, providing a high-quality image foundation for subsequent defect recognition. It truly realizes the advantages of dual-module coverage of the entire circumference, low cost and small image distortion, and solves the core pain points of the prior art, such as "high cost of multi-module detection, difficult debugging, and image distortion affecting recognition".

[0016] 2. Addressing the shortcomings of existing steel wire inspection methods, such as numerous surface impurities and high conveying vibration leading to low inspection accuracy, this equipment integrates a collaborative structure of annular cleaning nozzles and a crossbeam. The annular cleaning nozzles are coaxially arranged with the steel wire to be inspected, and a 360° uniform spray of clean airflow delivered by an external gas source effectively removes impurities such as wire drawing oil and metal debris from the steel wire surface, avoiding interference from impurities. At the same time, the crossbeam 4 is located upstream of the inspection unit and limits the steel wire, effectively suppressing lateral vibration during conveying and ensuring the stability of the steel wire's posture. This "cleaning + limiting" design provides a clean inspection surface for optical inspection and ensures the positional stability of the steel wire during image acquisition. It solves the problems of "numerous surface impurities and high steel wire vibration leading to missed or false detections" in existing technologies, achieving the advantages of "thorough cleaning, stable steel wire posture, and high inspection accuracy."

[0017] 3. Addressing the shortcomings of existing single-vision modules, such as limited field of view, high defect false negative rate, and poor equipment adaptability, this device features dual-vision ends in both the first and second vision modules. Combined with right-angle prisms for blind spot compensation, each vision module can directly capture the observable area while also acquiring the blind spot image through the prism. The two modules' detection areas complement each other, completely eliminating the visual blind spots of a single module and significantly reducing the defect false negative rate. Furthermore, the first and second reflecting prisms are standardized right-angle prisms, with a simple and easy-to-maintain mounting bracket structure. Adapting the device to different diameter steel wires only requires minor adjustments to the module spacing, without replacing core components, significantly reducing maintenance costs and adaptation difficulty. It truly achieves the advantages of "thorough blind spot compensation with dual vision ends, strong equipment adaptability, and easy maintenance," solving the pain points of "high false negative rate with single modules, poor equipment adaptability, and complex maintenance" in existing technologies. Attached Figure Description

[0018] Figure 1 This is a frontal perspective three-dimensional schematic diagram of the main structure of the present invention; Figure 2 This is a three-dimensional cross-sectional view of the main structure of the present invention; Figure 3 This is a partial three-dimensional schematic diagram of the main structure of the present invention; Figure 4 This is a three-dimensional schematic diagram of the cooperation relationship between the first vision module and the first reflecting prism of the present invention; Figure 5 This is a planar schematic diagram of the detection unit of the present invention when detecting the steel wire to be tested; Figure 6 This is a rear-view perspective view of the main structure of the present invention; Figure 7 For the present invention Figure 6 Enlarged 3D structural diagram at point A; Figure 8 This is a three-dimensional schematic diagram of the relationship between the annular cleaning nozzle and the airflow delivery pipe of the present invention; Figure 9 This is a flowchart of the detection method of the present invention.

[0019] In the diagram: 1. Equipment base; 2. Steel wire to be tested; 31. First optical detection module; 311. First vision module; 312. First reflecting prism; 32. Second optical detection module; 321. Second vision module; 322. Second reflecting prism; 33. Mounting bracket; 4. Crossbeam; 41. Annular cleaning nozzle; 42. Airflow delivery pipe. Detailed Implementation

[0020] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0021] It should be noted that the steel wire 2 to be tested is only used as the object of testing. Its material, diameter, etc. are determined according to the actual wire drawing process. The first optical detection module 31, including the first vision module 311 and the first reflecting prism 312, only provides optical detection functions: the first vision module 311 realizes image acquisition through the existing line scan camera structure, and the first reflecting prism 312 realizes optical path reversal through the existing right-angle triangular prism structure to fill the visual blind spot. The second optical detection module 32, including the second vision module 321 and the second reflecting prism 322, has the same function as the first optical detection module 31. The detection area is complementary through symmetrical layout.

[0022] The working principles (such as the imaging principle of a line scan camera, the total internal reflection principle of a right-angle prism, etc.) and specific structures (such as the resolution selection and frame rate parameters of the first vision module 311, the coating material and installation angle of the first reflection prism 312, etc.) of the above components are all existing technologies. Given the universality of these structures, their specific principles will not be elaborated further.

[0023] Example 1, as Figure 1 As shown, the image fusion-based steel wire drawing surface defect detection device includes a device base 1 and a detection unit arranged along the conveying path of the steel wire 2 to be detected. An external traction module is set along the conveying path of the steel wire 2 to be detected to provide power for the continuous conveying of the steel wire 2 to be detected. The detection unit includes a first optical detection module 31 and a second optical detection module 32. The first optical detection module 31 and the second optical detection module 32 are staggered along the axial direction of the steel wire 2 to be detected. The first optical detection module 31 is used to perform visual detection on half of the circumference area of ​​the steel wire 2 to be detected, and the second optical detection module 32 is used to perform visual detection on the other half of the circumference area of ​​the steel wire 2 to be detected.

[0024] like Figure 2 and Figure 3 As shown, the first optical detection module 31 and the second optical detection module 32 are both mounted on the equipment base 1 and are arranged in a staggered manner along the steel wire conveying path of the equipment base 1. The first optical detection module 31 includes a first vision module 311 and a first reflecting prism 312. The first reflecting prism 312 directs the surface reflected light of the steel wire portion that cannot be directly observed by the first vision module 311 to the first vision module 311, so that the first vision module 311 can acquire the reflected image. like Figure 5 As shown, the second optical detection module 32 includes a second vision module 321 and a second reflecting prism 322. The second reflecting prism 322 directs the surface reflected light of other parts of the wire that cannot be directly observed by the second vision module 321 to the second vision module 321, so that the second vision module 321 can acquire the reflected image and complement the detection area of ​​the first vision module 311.

[0025] It should be noted that both the first reflecting prism 312 and the second reflecting prism 322 are right-angle triangular prisms with anti-reflection coatings on their reflective surfaces. The first vision module 311 has two vision ends: one for directly capturing the observable area of ​​the steel wire 2 to be tested, and the other for collecting the blind area of ​​the steel wire after reflection by the first reflecting prism 312. The second vision module 321 has two corresponding vision ends: one for directly capturing the other observable area of ​​the steel wire 2 to be tested, and the other for collecting the blind area of ​​the steel wire after reflection by the second reflecting prism 322. The device base 1 is equipped with a mounting bracket 33, and both the first reflecting prism 312 and the second reflecting prism 322 are fixedly connected to the mounting bracket 33.

[0026] like Figure 6 to Figure 8 As shown, a crossbeam 4 is mounted on the equipment base 1. The crossbeam 4 is located on the side away from the first optical detection module 31 and the second optical detection module 32. The steel wire 2 to be tested passes through the crossbeam 4. An annular cleaning nozzle 41 is provided on the crossbeam 4, and the annular cleaning nozzle 41 is coaxially arranged with the steel wire 2 to be tested. An airflow delivery pipe 42 is connected to the outside of the annular cleaning nozzle 41. The other end of the airflow delivery pipe 42 is connected to an external gas source to provide cleaning airflow to the annular cleaning nozzle 41.

[0027] Specifically, after the external traction module is started, it drives the steel wire 2 to be tested to move at a constant speed along the preset conveying path of the equipment base 1. The steel wire 2 to be tested first passes through the crossbeam 4 assembled on the equipment base 1. The crossbeam 4 is clearly located on the side away from the first optical detection module 31 and the second optical detection module 32. This layout can avoid the crossbeam and related components from interfering with the optical acquisition of the detection module. At the same time, the steel wire 2 to be tested passes through the crossbeam 4 along the axis. The structural design of the crossbeam 4 can form a certain limiting effect on the steel wire, further limiting the lateral vibration of the steel wire 2 to be tested during the conveying process, and ensuring the stability of the steel wire posture during subsequent testing.

[0028] The annular cleaning nozzle 41 on the crossbeam 4 is arranged coaxially with the steel wire 2 to be tested. The annular cleaning nozzle 41 is connected to an external gas source through an air supply pipe 42. After the external gas source supplies cleaning air to the annular cleaning nozzle 41, the annular cleaning nozzle 41 can spray air evenly along the circumference of the steel wire 2 to be tested, effectively removing impurities such as wire drawing oil and metal debris attached to the surface of the steel wire 2 to be tested, ensuring that the surface of the steel wire is clean, avoiding impurities from obstructing or interfering with the image acquisition of the subsequent optical detection module, and laying the foundation for accurate detection.

[0029] The cleaned steel wire 2 continues to move toward the detection unit. The first optical detection module 31 and the second optical detection module 32 are both mounted on the equipment base 1 and are arranged in a staggered manner along the steel wire conveying path of the equipment base 1. The mounting bracket 33 mounted on the surface of the equipment base stably fixes the first reflecting prism 312 and the second reflecting prism 322 to ensure that the prism position is accurate and not easily shifted.

[0030] The first vision module 311 of the first optical detection module 31 has two vision ends. One vision end can directly capture the observable area of ​​the steel wire 2 to be detected. However, due to the cylindrical structure of the steel wire, the vision end cannot cover half of the circumference of the steel wire 2 to be detected, resulting in a visual blind spot. At this time, the first reflecting prism 312 plays a role in directing the surface reflected light of the blind area of ​​the steel wire that the first vision module 311 cannot directly observe to the other vision end of the first vision module 311, so that the first vision module 311 can completely acquire the image of the first semicircular area of ​​the steel wire 2 to be detected.

[0031] Similarly, the second vision module 321 of the second optical detection module 32 also has two vision ends. One vision end directly captures the image of another observable area of ​​the steel wire 2 to be detected, while the second reflecting prism 322 directs the surface reflected light from other blind areas of the steel wire that cannot be directly observed by the second vision module 321 to the other vision end of the second vision module 321, enabling the second vision module 321 to acquire the image of the second semi-circular area of ​​the steel wire 2 to be detected. Since the detection areas of the first vision module 311 and the second vision module 321 are complementary, the full-circumference visual detection of the steel wire 2 to be detected is finally achieved through the collaborative work of the first optical detection module 31 and the second optical detection module 32.

[0032] Example 2, as Figure 9 As shown, the method for detecting surface defects of steel wire drawing based on image fusion includes the following steps: S1: Start the external traction module, set the conveying speed threshold of the steel wire 2 to be detected, and simultaneously calibrate the acquisition parameters of the first optical detection module 31 and the second optical detection module 32—adjust the misalignment distance between the two detection modules according to the diameter of the steel wire 2 to be detected, and calibrate the image acquisition frequency; S2: The external traction module drives the steel wire 2 to be tested to move at a constant speed along the conveying path, while simultaneously triggering the first optical detection module 31 and the second optical detection module 32 to work synchronously. The first optical detection module 31 acquires the direct observation image and the prism reflection-completed image of the first semi-circular area of ​​the steel wire 2 to be detected; The second optical detection module 32 acquires the direct observation image and the prism reflection-completed image of the second semi-circular region of the steel wire 2 to be detected; The acquisition process of the two detection modules is linked by the synchronous controller on the equipment base 1 to ensure that the images acquired at the same time correspond to the same axial position of the steel wire 2 to be detected; S3: Process the image acquired in S2: Use an adaptive median filtering algorithm to specifically eliminate image noise caused by airflow disturbances and slight vibrations of the steel wire, while preserving defect details; Illumination compensation: By stretching grayscale, local uneven illumination caused by differences in the light path during prism reflection is compensated, so that the grayscale distribution of the image is uniform. Defect Enhancement: Apply edge enhancement algorithms to highlight the edge contours of minute defects on the steel wire surface, thereby improving the sensitivity of subsequent defect identification; Coordinate correction: Based on the preset misalignment distance between the two detection modules, the coordinate mapping algorithm is used to eliminate the axial offset of the image caused by the misalignment distribution along the axis, ensuring that the axial coordinates of the two types of images are accurately aligned. S4: Based on the cylindrical surface geometric features of the steel wire 2 to be detected, establish an image coordinate mapping model: extract the edge features of the directly observed area in the image of the first optical detection module 31 and the texture features of the reflection completion area in the image of the second optical detection module 32; A combination of weighted average fusion and gradient domain fusion is used to stitch the preprocessed images of the detection unit into a fully unfolded image of the steel wire 2 to be detected, ensuring that there are no obvious gray-scale breaks at the stitching points. S5: The model first locates the suspected defect area, and then extracts the geometric and texture features of the defect; Suspected defects are classified using a pre-defined defect classifier, and the defect type and confidence level are output. S6: Record the specific location information of each defect and generate an inspection report; at the same time, feed the defect data back to the control system of the wire drawing equipment to adjust the drawing die parameters or lubrication dosage to reduce the occurrence of subsequent defects.

[0033] It should be noted that, in this document, relational terms such as "first" and "second" are used only 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 apparatus 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 apparatus. 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 apparatus that includes said element.

[0034] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A surface defect detection device for steel wire drawing based on image fusion, comprising a device base (1) and a detection unit arranged along the conveying path of the steel wire (2) to be detected, wherein an external traction module is set along the conveying path of the steel wire (2) to be detected, for providing power for the continuous conveying of the steel wire (2), characterized in that: The detection unit includes a first optical detection module (31) and a second optical detection module (32). The first optical detection module (31) and the second optical detection module (32) are staggered along the axial direction of the steel wire (2) to be detected. The first optical detection module (31) is used to perform visual detection on half of the circumference of the steel wire (2) to be detected, and the second optical detection module (32) is used to perform visual detection on the other half of the circumference of the steel wire (2) to be detected.

2. The image fusion-based steel wire drawing surface defect detection device according to claim 1, characterized in that: The first optical detection module (31) and the second optical detection module (32) are both mounted on the equipment base (1) and are arranged in a staggered manner along the wire conveying path of the equipment base (1); The first optical detection module (31) includes a first vision module (311) and a first reflecting prism (312). The first reflecting prism (312) directs the surface reflected light of the steel wire portion area that cannot be directly observed by the first vision module (311) to the first vision module (311), so that the first vision module (311) can acquire the reflected image. The second optical detection module (32) includes a second vision module (321) and a second reflection prism (322). The second reflection prism (322) directs the surface reflected light of other parts of the wire that cannot be directly observed by the second vision module (321) to the second vision module (321), so that the second vision module (321) can acquire the reflected image and complement the detection area of ​​the first vision module (311).

3. The image fusion-based steel wire drawing surface defect detection device according to claim 2, characterized in that: Both the first reflecting prism (312) and the second reflecting prism (322) are right-angle triangular prisms, and the reflecting surfaces are coated with an anti-reflection film.

4. The image fusion-based steel wire drawing surface defect detection device according to claim 2, characterized in that: The first vision module (311) is provided with two vision ends: one for directly capturing the observable area of ​​the steel wire (2) to be detected, and the other for collecting the blind area of ​​the steel wire after being reflected by the first reflecting prism (312); The second vision module (321) is provided with two vision ends: one for directly capturing the image of another observable area of ​​the steel wire (2) to be detected, and the other for collecting the image of the blind area of ​​the steel wire after being reflected by the second reflecting prism (322).

5. The image fusion-based steel wire drawing surface defect detection device according to claim 2, characterized in that: The device base (1) is equipped with a mounting bracket (33), and the first reflecting prism (312) and the second reflecting prism (322) are both fixedly connected to the mounting bracket (33).

6. The image fusion-based steel wire drawing surface defect detection device according to claim 1, characterized in that: The equipment base (1) is equipped with a crossbeam (4), which is located on the side away from the first optical detection module (31) and the second optical detection module (32). The steel wire (2) to be tested passes through the crossbeam (4), and an annular cleaning nozzle (41) is provided on the crossbeam (4). The annular cleaning nozzle (41) is coaxially arranged with the steel wire (2) to be tested.

7. The image fusion-based steel wire drawing surface defect detection device according to claim 6, characterized in that: The annular cleaning nozzle (41) is externally connected to an airflow delivery pipe (42), and the other end of the airflow delivery pipe (42) is connected to an external gas source to provide cleaning airflow to the annular cleaning nozzle (41).

8. A method for detecting surface defects in steel wire drawing based on image fusion, applied to the image fusion-based steel wire drawing surface defect detection equipment described in claims 3-5, characterized in that: Includes the following steps: S1: Start the external traction module, set the conveying speed threshold of the steel wire (2) to be tested, and simultaneously calibrate the acquisition parameters of the first optical detection module (31) and the second optical detection module (32) - adjust the misalignment distance of the two detection modules according to the diameter of the steel wire (2) to be tested, and calibrate the image acquisition frequency; S2: The external traction module drives the steel wire (2) to be tested to move at a constant speed along the conveying path, and at the same time triggers the first optical detection module (31) and the second optical detection module (32) to work synchronously: The first optical detection module (31) acquires the direct observation image and the prism reflection-completed image of the first semicircular area of ​​the steel wire to be detected (2); The second optical detection module (32) acquires the direct observation image and the prism reflection-completed image of the second semicircular area of ​​the steel wire (2) to be detected; The acquisition process of the two detection modules is linked by the synchronous controller on the equipment base (1) to ensure that the images acquired at the same time correspond to the same axial position of the steel wire (2) to be detected; S3: Process the image acquired in S2: An adaptive median filtering algorithm is used to specifically eliminate image noise caused by airflow disturbances and slight vibrations of the steel wire, while preserving defect details. Illumination compensation: By stretching grayscale, local uneven illumination caused by differences in the light path during prism reflection is compensated, so that the grayscale distribution of the image is uniform. Defect Enhancement: Apply edge enhancement algorithms to highlight the edge contours of minute defects on the steel wire surface, thereby improving the sensitivity of subsequent defect identification; Coordinate correction: Based on the preset misalignment distance between the two detection modules, the coordinate mapping algorithm is used to eliminate the axial offset of the image caused by the misalignment distribution along the axis, ensuring that the axial coordinates of the two types of images are accurately aligned. S4: Based on the cylindrical surface geometric features of the steel wire (2) to be detected, establish an image coordinate mapping model: extract the edge features of the directly observed area in the image of the first optical detection module (31) and the texture features of the reflection completion area in the image of the second optical detection module (32); By combining weighted average fusion and gradient domain fusion, the preprocessed image of the detection unit is stitched together to form a full-circumference unfolded image of the steel wire to be detected (2), ensuring that there is no obvious gray-scale discontinuity at the stitching point; S5: The model first locates the suspected defect area, and then extracts the geometric and texture features of the defect; Suspected defects are classified using a pre-defined defect classifier, and the defect type and confidence level are output. S6: Record the specific location information of each defect and generate an inspection report; at the same time, feed the defect data back to the control system of the wire drawing equipment to adjust the drawing die parameters or lubrication dosage to reduce the occurrence of subsequent defects.

Images