Dual torsion spring imaging detection method and detection device

Abstract

This invention belongs to the field of spring visual inspection technology and discloses a double torsion spring imaging inspection method. The specific steps are as follows: Step 1: Position the double torsion spring to be inspected at the feed position of the turntable; Step 2: Drive the double torsion spring through three image acquisition stations in sequence by rotating the turntable, and collect images of the double torsion spring in different postures; Step 3: Analyze and process the multi-posture images to automatically determine whether the forming quality of the double torsion spring is qualified; Step 4: Based on the determination result, the qualified double torsion spring and the unqualified double torsion spring are output separately from different discharge positions. This invention adopts a three-station differentiated perspective collaborative imaging scheme, and through the acquisition of front, side, and oblique three-dimensional multi-posture images, it simultaneously covers the planar size parameters and three-dimensional deformation state of the double torsion spring, completely solving the problems of blind spots, inability to identify three-dimensional torsional deviations, asymmetrical deformations, and other hidden defects in traditional single-view visual inspection.

Description

Technical Field

[0001] This invention belongs to the field of spring visual inspection technology, specifically a double torsion spring imaging inspection method and inspection equipment. Background Technology

[0002] Double torsion springs are commonly used elastic connectors in various precision machinery, home appliance parts, automotive hardware, and intelligent equipment. They rely on the elastic deformation of the double torsion arms to achieve functions such as reset, clamping, and buffering. Their forming accuracy, structural symmetry, and dimensional consistency directly affect the operational stability and service life of the entire equipment. During the production and processing of double torsion springs, defects such as deviations in torsion arm length, torsion angle offsets, spring deformation, burrs on edges and corners, and asymmetrical forming are very likely to occur. If unqualified products flow into the assembly process, it will lead to quality problems such as equipment jamming, reset failure, and abnormal noise.

[0003] Currently, the industry primarily relies on manual visual sampling for the inspection of double torsion springs, supplemented by single-angle machine vision inspection. Manual inspection depends on operator experience, resulting in low efficiency, significant human fatigue errors, high rates of missed and false detections, and inconsistent inspection standards, making it unsuitable for large-scale mass production. Existing machine vision inspection equipment mostly uses single-station, single-angle imaging, capturing only the appearance information of a single posture of the double torsion spring. Because double torsion springs have a three-dimensional, irregular structure, a single perspective cannot fully cover the key inspection areas of the torsion arm, spring body, and connecting ends, making it difficult to identify hidden defects such as multi-angle deformation and symmetry deviations. The inspection accuracy and comprehensiveness are insufficient, and there is a lack of an integrated structure for automated flow and precise sorting, leading to a disconnect between inspection and sorting processes and a low overall level of automation. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to provide a dual torsion spring imaging detection method and detection device to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a dual torsion spring imaging detection method, the specific steps of which are as follows: Step 1: Position the double torsion spring to be tested at the feed position of the turntable; Step Two: The rotation of the turntable drives the double torsion spring through three image acquisition stations in sequence, and collects images of the double torsion spring in different postures. Step 3: Multi-pose images are analyzed and processed to automatically determine whether the molding quality of the double torsion spring is up to standard; Step Four: Based on the judgment results, qualified double torsion springs and unqualified double torsion springs are output separately from different outlets.

[0006] As a preferred technical solution of the present invention, the turntable in step two rotates intermittently, and the double torsion springs are precisely stopped for a preset time at each image acquisition station to complete steady-state image acquisition.

[0007] As a preferred technical solution of the present invention, the three image acquisition stations in step two correspond to the frontal posture, side posture, and oblique three-dimensional posture of the double torsion spring, respectively, to realize full-dimensional image acquisition of the double torsion spring.

[0008] As a preferred technical solution of the present invention, the image analysis and processing in step three includes image noise reduction, grayscale processing, contour extraction, standard model comparison, geometric tolerance detection, and automatic identification of double torsion spring size deviation, deformation, burrs, and asymmetric defects.

[0009] To achieve the other objective mentioned above, the present invention provides the following technical solution: a double torsion spring imaging detection device, comprising a turntable, wherein the turntable is provided with a feeding position for placing the double torsion spring to be detected; three image acquisition devices for acquiring images of the double torsion spring in different postures are arranged sequentially along the rotation direction of the turntable; and two discharge positions are provided downstream of the three image acquisition devices for outputting qualified and unqualified double torsion springs respectively.

[0010] As a preferred embodiment of the present invention, the feed position of the turntable is provided with a contour positioning fixture that matches the shape of the double torsion spring, which is used to limit and fix the double torsion spring to be tested.

[0011] As a preferred embodiment of the present invention, the image acquisition device is equipped with a ring-shaped shadowless supplementary light source, and the supplementary light source is coaxially arranged with the image acquisition device.

[0012] As a preferred embodiment of the present invention, it further includes an image processing controller, which is electrically connected to three image acquisition devices respectively, and is used to receive image data, analyze and determine the quality of workpieces and output sorting control signals.

[0013] As a preferred embodiment of the present invention, each of the two discharge positions is equipped with a pneumatic sorting and pushing mechanism, which receives a signal from the image processing controller to complete a precise sorting action.

[0014] The beneficial effects of this invention are as follows: 1. This invention adopts a three-station differentiated perspective collaborative imaging scheme, which simultaneously covers the planar dimensional parameters of the double torsion spring and the three-dimensional deformation state through frontal, lateral, and oblique three-dimensional multi-pose image acquisition methods. This completely solves the problems of blind spots, inability to identify three-dimensional torsional deviations, asymmetric deformations, and other hidden defects in traditional single-view visual inspection. Combined with ring shadowless supplementary lighting, steady-state parking imaging, and standardized image algorithm processing, it effectively avoids interference factors such as metal reflection, imaging distortion, and subjective judgment errors. It can accurately identify various forming defects such as dimensional deviations, burrs, deformations, and structural asymmetry, thereby improving the consistency of inspection and the defect identification coverage.

[0015] 2. This invention relies on an intermittent indexing turntable to achieve fully unmanned closed-loop operation of automatic workpiece feeding, precise station switching, steady-state imaging detection, intelligent judgment, and automatic sorting and unloading. The processes at each station are connected in parallel without interference, which gets rid of the drawbacks of low efficiency and disjointed processes in traditional manual sampling inspection. The equipment can realize large-scale, continuous production line inspection operations with stable inspection cycle and fast response speed, reducing the inspection time of a single piece and effectively improving the overall operation efficiency of mass production inspection of double torsion springs, making it suitable for large-scale industrial production scenarios.

[0016] 3. This invention adopts an integrated turntable multi-station layout, with a simple and compact overall structure, low equipment failure rate, and stable operation. The contour positioning fixture ensures uniform workpiece placement posture, and the adjustable imaging support structure can adapt to the detection needs of double torsion springs of different specifications and sizes, making it widely applicable. At the same time, relying on the integrated electronic control intelligent judgment and pneumatic precision sorting structure, it realizes precise linkage of detection, judgment, and sorting, with high sorting accuracy. No manual intervention is required throughout the process, effectively reducing labor costs and the probability of human error. Attached Figure Description

[0017] Figure 1 This is a flowchart of the dual torsion spring imaging detection method of the present invention; Figure 2 This is a schematic diagram of the dual torsion spring imaging detection device of the present invention. Detailed Implementation

[0018] 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 scope of protection of the present invention.

[0019] like Figure 1 As shown, this embodiment of the invention provides a dual-torsion spring imaging detection method, the specific steps of which are as follows: Step 1: Position the double torsion spring to be tested at the feed position of the turntable; Step Two: The rotation of the turntable drives the double torsion spring through three image acquisition stations in sequence, and collects images of the double torsion spring in different postures. Step 3: Multi-pose images are analyzed and processed to automatically determine whether the molding quality of the double torsion spring is up to standard; Step Four: Based on the judgment results, qualified double torsion springs and unqualified double torsion springs are output separately from different outlets.

[0020] This solution uses a turntable to carry the workpieces and, in conjunction with a three-station step-by-step image acquisition method, can sequentially acquire imaging information of multiple sets of double torsion springs in different postures. By comprehensively analyzing the multi-dimensional images, the quality of the workpieces is determined, and finally, based on the determination results, good and bad products are automatically separated and discharged. The entire process forms a continuous and closed-loop automated inspection procedure, effectively covering all the inspection points of the double torsion spring three-dimensional structure.

[0021] In step two, the turntable rotates intermittently, and the double torsion springs are precisely stopped for a preset time at each image acquisition station to complete steady-state image acquisition.

[0022] Precise intermittent rotation and steady-state parking are achieved through the coordinated operation of a stepper motor and an indexing control program. Specifically, the equipment uses a high-precision stepper motor as the turntable drive source. The motor is connected to a programmable controller (PLC), which pre-sets the rotation angle, rotation speed, and parking duration to match the station spacing. During operation, the controller outputs pulse signals to control the stepper motor to rotate quantitatively, driving the turntable to complete a fixed-angle indexing jump, precisely displacing the workpiece carrying a double torsion spring to the corresponding image acquisition station. After the turntable completes the jump, the motor is instantly locked and braked, keeping the turntable absolutely stationary. The workpiece maintains a preset parking duration of 0.2s-0.5s at the station. During this parking window, the image acquisition device is triggered to capture images, completely avoiding displacement deviations caused by the dynamic rotation of the turntable. This ensures that each image acquisition is completed under steady-state conditions, guaranteeing images without ghosting, distortion, or high positional consistency.

[0023] In step two, the three image acquisition stations correspond to the frontal, side, and oblique three-dimensional postures of the double torsion spring, respectively, to achieve full-dimensional image acquisition of the double torsion spring.

[0024] Three image acquisition stations employ fixed, differentiated installation angles in conjunction with a stationary turntable to achieve precise image acquisition. The specific acquisition methods are as follows: Frontal image: The image acquisition device is vertically positioned facing the upper surface of the double torsion spring, directly aligned with the center of the spring body, to acquire a frontal planar image of the workpiece, used to collect planar dimensional data such as torsion arm length, left-right spacing, and planar contour; Side image: The image acquisition device is horizontally positioned laterally, parallel to the extension direction of the double torsion spring's torsion arm, to acquire a side projection image of the workpiece, used to collect side detail data such as torsion arm thickness, end face flatness, and side burrs; Oblique 3D image: The image acquisition device is positioned at a 45° oblique angle, tilted towards the overall structure of the double torsion spring, to acquire a 3D perspective image of the workpiece, capturing three-dimensional morphological information such as the torsion arm's torsion angle, the spring body's curvature, and the symmetrical state of the two sides of the torsion arms. The three stations acquire images independently and in shifts, with complementary data, fully covering the planar and 3D structure of the double torsion spring, achieving blind-spot-free, full-dimensional image acquisition. The image acquisition device specifically consists of a high-frame-rate industrial monochrome camera paired with a fixed-focus high-definition lens. The camera resolution is no less than 2 million pixels, and with the global shutter imaging mode, it can capture still images without motion blur or distortion within a short dwell time. The lens adopts a low-distortion fixed-focus lens, which effectively reduces edge imaging distortion errors and ensures the accuracy of dimensional measurement. The camera is locked in place by a fixed bracket, and the angle can be finely adjusted and locked in place to adapt to the shooting field of view requirements of different specifications of double torsion springs.

[0025] In step three, image analysis and processing includes image denoising, grayscale processing, contour extraction, standard model comparison, geometric tolerance detection, and automatic identification of double torsion spring dimensional deviations, deformations, burrs, and asymmetric defects.

[0026] Image noise reduction and grayscale processing remove interference from the image, and contour extraction accurately locates the workpiece's outer boundary. By comparing and analyzing pre-stored standard models and geometric tolerance thresholds, various forming defects such as dimensional deviations, overall deformation, edge burrs, and left-right structural asymmetry can be automatically identified. The specific implementation of image analysis and processing is as follows: Image noise reduction uses a Gaussian filtering algorithm to smooth the acquired original image, filtering out environmental noise and circuit pixel noise while retaining the effective contour information of the workpiece; Grayscale processing converts the color image into an 8-bit grayscale image, simplifying image moments. The system uses array data to reduce interference from metallic reflective colors, improving the speed and stability of subsequent calculations; contour extraction uses edge detection algorithms to extract the outer contour of the double torsion spring, the edge of the torsion arm, and the boundary coordinates of the spring body end face, accurately locking the effective detection area of ​​the workpiece and eliminating background interference areas; standard model comparison uses pre-entered contour templates and standard size parameter libraries of qualified standard double torsion springs, and performs pixel matching and contour overlap comparison between the real-time extracted workpiece contour and the standard template; and geometric tolerance detection uses the system to automatically calculate key parameters such as torsion arm length, spacing, torsion angle, and symmetry, and compare them with preset tolerance thresholds. After completing the standard model contour overlap matching and key shape and position parameter calculation, the system enters a hierarchical image comparison and judgment process, which is divided into four progressive judgment stages: coarse matching initial screening, multi-dimensional parameter quantitative comparison, defect feature secondary verification, and comprehensive weighted quality judgment. Each stage completes defect identification and workpiece quality grading. The detailed judgment process is as follows: I. Coarse Matching Initial Screening: Preliminary Judgment of Global Contour Overlap Threshold The system first calculates the global pixel overlap rate between the real-time workpiece contour and the pre-stored standard template, and presets a first-level coarse overlap threshold (the baseline qualified overlap threshold is 95%). If the contour overlap rate is <80%, the workpiece is directly judged to have major defects such as severe overall deformation, large misalignment of torsion arm, and twisting of spring body. The subsequent fine parameter comparison process is skipped and the workpiece is directly marked as unqualified. If the contour overlap rate is between 80% and 95%, it is marked as a suspected defective workpiece and enters the next level of refined parameter quantitative comparison process; If the contour overlap rate is ≥95%, it is determined that the overall shape of the workpiece has no obvious deviation, and only quantitative verification of local dimensions and minor features is required. This step quickly filters out severely defective workpieces, reduces the amount of high-precision calculations required in the background, and improves the overall inspection cycle time.

[0027] II. Quantitative Comparison of Multi-Dimensional Parameters: Item Tolerance Threshold Verification The system retrieves all key dimensions and upper and lower limits of geometric tolerances for double torsion springs from the pre-stored database, and compares and verifies each of the calculated real-time parameters. The judgment rules for each item are as follows: Torsion arm length comparison judgment: Extract the effective extension length values ​​of the left and right torsion arms respectively, and compare them with the standard length tolerance range: if the length of one side exceeds the upper / lower tolerance threshold, it is judged as a torsion arm length deviation defect; at the same time, calculate the difference in length between the left and right torsion arms, and if the difference exceeds the symmetry tolerance, it is judged as a length asymmetry defect. Comparison and judgment of the center distance of the torsion arm: Calculate the distance between the root and end of the double torsion arm and match the plane dimension tolerance: if the distance exceeds the tolerance at any point, the spring body opening size is determined to be unqualified and there is a molding offset problem. Torsion angle comparison and judgment: Extract the spatial torsion angle of the torsion arm by retrieving a 45° oblique stereo image and compare it with the standard torsion angle threshold: if the angle deviation exceeds the allowable range, a stereo torsion deviation defect is determined. This type of defect cannot be identified by relying solely on planar images and is a core verification item for three-view imaging. Spring symmetry comparison and judgment: Using the central axis of symmetry of the double torsion spring as a reference, calculate the lateral and longitudinal offsets of the left and right torsion arm endpoints, spring ring edge, and bending point respectively, and sum the overall symmetry deviation value. If the deviation exceeds the threshold, the structure is judged to be asymmetrical. Thickness and end face flatness comparison judgment: Based on horizontal side-view images, extract the torsion arm thickness and the height difference of the spring body end face. When the thickness is uneven or the end face concavity and convexity difference exceeds the limit, record the side forming defects. The system generates a separate verification mark for each parameter, distinguishes between three deviation levels: "slight deviation, moderate deviation, and severe deviation," and stores the deviation values ​​simultaneously for subsequent comprehensive judgment.

[0028] III. Secondary Verification of Defect Features: Identification of Local Contour Imperfections After the overall dimensions are verified, local pixel feature retrieval is performed on the workpiece contour edges, corners, and torsion arm side areas to accurately identify subtle hidden defects: Burr defect assessment: The edge protrusion pixel clusters are retrieved pixel by pixel along the complete outline of the workpiece, and the area and height of the protrusion area are counted. When the size of the local edge protrusion exceeds the allowable threshold for burrs, the corner burr defect is determined. At the same time, two defect levels are distinguished: micro burrs and large-area continuous burrs. Local deformation / notch identification: Search for concave pixel areas within the contour. If regular / irregular concave notches or local bending deformations are found, record the local forming defects and simultaneously match the notch position and defect size parameters. Surface defect auxiliary inspection: Based on the difference between bright and dark pixels in grayscale images, identify grayscale abrupt change areas such as indentations and scratches on the surface of the spring. When the area of ​​the abrupt change area exceeds the limit value, it is simultaneously included in the defect record. This step provides a fallback for identifying minute, hidden defects that coarse matching and size comparison cannot detect, eliminating the risk of missing these minor defects.

[0029] IV. Comprehensive Weighted Quality Assessment: Output the assessment results by summarizing the defect levels. The system integrates all defect markers, defect quantities, and defect severity levels generated from previous comparisons and verifications, and completes the final quality assessment through preset weighted judgment logic: Pass / Fail Judgment Rules: There are no out-of-tolerance parameters, no burrs / deformation / notches or other local defects, and all parameters such as contour overlap, size, symmetry and torsion angle are within the standard tolerance range. The weighted defect score is 0, and the "qualified" judgment signal is output directly. Non-compliance determination falls into three categories: If any serious out-of-tolerance / major deformation defect exists (contour overlap rate less than 80%, torsion arm length deviation exceeding the upper limit by 50%, torsion angle deviation significantly, etc.), it will be directly judged as unqualified; No major defects, but two or more moderate dimensional deviations, or the presence of large areas of burrs or obvious asymmetrical deformation, shall be deemed unqualified. If there is only a single, slight burr or minor dimensional deviation, but it exceeds the factory's allowable tolerance standard, it is deemed unqualified. Detection signal synchronous output: After completing the comprehensive judgment, the image processing controller synchronously stores all images of the workpiece, various comparison parameters, and defect location records, and transmits corresponding control signals to the downstream pneumatic sorting and pushing mechanism: qualified workpieces transmit no action instructions, and unqualified workpieces push sorting trigger instructions, providing a judgment basis for automated diversion and sorting.

[0030] V. Comparison of data archiving and dynamic template correction auxiliary logic After a single image comparison and judgment process is completed, the system automatically archives all comparison data, defect screenshots, and parameter deviation values ​​for this workpiece. At the same time, it performs long-term statistical analysis on the distribution patterns of defects in batch workpieces. When workpieces of the same specification continuously exhibit the same type of systematic dimensional deviation, it can prompt operators to fine-tune the tolerance threshold of the standard model template, the supplementary lighting parameters, or the imaging shooting angle, thereby achieving dynamic optimization of the detection comparison standard and continuously reducing the probability of false detection and missed detection in batch detection.

[0031] like Figure 2 As shown, this embodiment of the invention also provides a double torsion spring imaging detection device, including a turntable, on which a feeding position for placing the double torsion spring to be detected is provided; three image acquisition devices for acquiring images of the double torsion spring in different postures are arranged sequentially along the rotation direction of the turntable; downstream of the three image acquisition devices are two discharge positions for outputting qualified and unqualified double torsion springs respectively.

[0032] Relying on the turntable as the workpiece conveyor, and utilizing the multi-station structure arranged circumferentially on the turntable, the workstations for workpiece loading, multi-angle imaging detection, and sorting and unloading are switched sequentially. The three imaging devices can capture the appearance of the double torsion springs from different angles. The downstream dual unloading positions can match the detection results, enabling independent partitioning and unqualified workpiece unloading. The structural layout is adapted to continuous batch inspection operations.

[0033] The turntable's feed position is equipped with a contour positioning fixture that matches the shape of the double torsion spring, used to limit and fix the double torsion spring to be tested.

[0034] The contour positioning fixture is customized with a limiting cavity based on the shape and structure of the double torsion spring. It can limit and fix the double torsion spring placed at the feeding position in all directions, restricting the workpiece from shifting, flipping, or sliding during turntable rotation, station switching, and imaging, and ensuring that the placement posture and position of each batch of workpieces are consistent.

[0035] The image acquisition device is equipped with a ring-shaped shadowless supplementary light source, which is coaxially arranged with the image acquisition device.

[0036] The ring-shaped shadowless supplementary light source is coaxially assembled with the image acquisition device, which can form a uniform and blind-angle-free illumination field above the workpiece. Taking into account the characteristics of metal double torsion springs that are prone to reflection and shadows, it weakens local light spots and dark area defects, improves the imaging clarity of workpiece contours and edge details, and provides high-quality original images for subsequent image algorithm analysis.

[0037] It also includes an image processing controller, which is electrically connected to three image acquisition devices to receive image data, analyze and determine the quality of workpieces, and output sorting control signals.

[0038] The image processing controller establishes a signal connection with three sets of image acquisition devices, and can synchronously receive image data collected from multiple workstations. It can centrally complete image calculation, parameter comparison and quality judgment. At the same time, the controller can output corresponding switch control signals according to the judgment results, providing a precise source of instructions for the subsequent sorting mechanism's sorting actions.

[0039] Both discharge positions are equipped with pneumatic sorting and pushing mechanisms, which receive signals from the image processing controller to complete precise sorting actions.

[0040] The pneumatic sorting and pushing mechanism is set on the feeding side of the two sets of discharge positions. It can receive sorting signals from the image processing controller in real time. For workpieces that are judged to be qualified or unqualified, the corresponding pushing mechanism will precisely extend and retract to push the workpiece to the corresponding discharge position, so as to achieve precise linkage and matching between the detection results and the sorting and discharge actions.

[0041] Working principle and usage process: By adopting a rotary intermittent multi-station flow structure combined with multi-view machine vision imaging technology, a fully automated closed-loop detection process is achieved for the dual torsion spring, from material feeding and positioning, multi-posture steady-state image acquisition, intelligent image analysis, automatic quality judgment to precise automatic sorting and unloading. The core hardware of the overall equipment includes an intermittent indexing rotary table, a contour positioning fixture, three sets of differentiated image acquisition devices, a coaxial ring shadowless supplementary light source, an image processing controller, and a dual-station pneumatic sorting and pushing mechanism.

[0042] In the workpiece loading and positioning process, the turntable of the equipment has multiple feeding positions arranged in a circular array. Each feeding position is matched with a customized contour positioning fixture that is perfectly adapted to the shape and structure of the double torsion spring. The fixture has a dedicated limiting cavity based on the contour dimensions of the double torsion spring body and the double torsion arms. It can provide all-round limiting constraints on the placed double torsion spring to be tested, avoiding problems such as slippage, offset, flipping, and posture misalignment of the workpiece during the high-speed indexing rotation of the turntable and the switching of work positions. This ensures that the placement position and posture of each double torsion spring to be tested are completely uniform in the initial work position.

[0043] After the workpiece is positioned, the programmable controller of the equipment, together with a high-precision stepper motor, drives the turntable to complete intermittent indexing rotation. The stepper motor receives precise pulse signals output by the controller, and each rotation angle strictly matches the circumferential distance between adjacent image acquisition stations, realizing quantitative and precise station jumps. When the turntable delivers the double torsion spring to be inspected to any image acquisition station, the stepper motor is instantly locked and braked, and the turntable comes to a complete stop. The workpiece maintains a preset steady-state dwell time of 0.2s to 0.5s at the corresponding imaging station. The entire image acquisition process is completed under the condition that the workpiece is absolutely stationary, ensuring the clarity, stability and consistency of each set of acquired images, and meeting the imaging requirements of high-precision dimensional inspection and defect identification.

[0044] The three image acquisition stations, arranged sequentially along the clockwise rotation direction of the turntable, employ a differentiated perspective layout to cooperate in completing blind-spot-free, full-dimensional image acquisition of the double torsion springs. All three image acquisition devices utilize high-frame-rate industrial monochrome cameras paired with low-distortion fixed-focus high-definition lenses and a global shutter imaging mode. This allows for rapid, distortion-free, and motion-free static image capture within a short dwell window, ensuring accurate dimensional measurements. The cameras are mounted on lockable, adjustable brackets, allowing for fine-tuning of the shooting angle and field of view according to the different dimensions of the double torsion springs. Furthermore, to accommodate the reflective and shadow-generating properties of the metal double torsion springs, each image acquisition device is coaxially equipped with a ring-shaped shadowless supplementary light source. This light source creates a uniform, blind-spot-free, soft illumination field in the workpiece inspection area, weakening bright spots on the metal surface, eliminating shadows in structural dark areas, and improving the imaging clarity of workpiece contours and fine structures. This provides high-quality raw image data for backend image algorithm processing.

[0045] During the imaging acquisition process, the first set of image acquisition devices vertically faces the upper surface and center of the double torsion spring, capturing a planar image of the workpiece's front view. This accurately obtains planar dimensional information such as the overall planar contour of the double torsion spring, the length of the two torsion arms, the distance between the torsion arms, the outer diameter of the spring body, and the flatness of the end face. The second set of image acquisition devices uses a 45° oblique angle to capture a three-dimensional image of the workpiece, capturing three-dimensional morphological data that cannot be obtained through planar photography, such as the actual torsion angle of the torsion arms, the degree of bending deformation of the spring body, and the three-dimensional symmetry of the two torsion arms. The third set of image acquisition devices is horizontally positioned laterally, parallel to the extension direction of the torsion arms, capturing a side view image of the workpiece. This accurately presents subtle defect information such as the uniformity of the torsion arm thickness, side burrs, end notches, and side flatness.

[0046] The multi-pose raw images acquired by the three image acquisition devices are transmitted synchronously in real time to the image processing controller. The controller's built-in image analysis algorithm performs fully automatic data processing and quality judgment: First, the raw images are denoised using a Gaussian filtering algorithm to filter out invalid interference information such as shooting environment noise and circuit pixel noise, while fully preserving the effective contour features of the workpiece; then, the color images are converted into 8-bit grayscale images to simplify the image operation matrix, weaken the color interference caused by metal reflection, and improve the algorithm's operation speed and stability; finally, a dedicated edge detection algorithm is used to accurately extract the outer contour of the double torsion spring, the edge of the torsion arm, and the spring body. Key coordinate information such as end face boundary is used to accurately divide the effective detection area and eliminate background interference areas. After image preprocessing, the real-time extracted workpiece contour and size parameters are matched with the standard qualified double torsion spring contour template and standard size parameter library stored in the equipment for pixel matching and contour overlap comparison. At the same time, combined with the preset form and position tolerance threshold, key form and position parameters such as torsion arm length deviation, spacing deviation, torsion angle error, and overall symmetry are automatically calculated. Finally, the workpiece is comprehensively judged to determine whether there are various forming defects such as dimensional deviation, overall deformation, edge burrs, and structural asymmetry, and the accurate judgment signal of qualified or unqualified is automatically output.

[0047] After completing single-workpiece image inspection and quality judgment, the turntable continues to perform intermittent indexing rotation, conveying the inspected workpieces to the sorting and unloading station downstream of the three sets of image acquisition devices. This station is equipped with independent qualified product unloading and unqualified product unloading stations. Each of the two unloading stations is equipped with an independent pneumatic sorting and pushing mechanism, and all pushing mechanisms are electrically connected to the image processing controller, which can receive judgment commands in real time and complete precise action responses. In the actual sorting process, for double torsion springs judged as qualified, the pneumatic pushing mechanism on the unqualified product side remains stationary, and qualified workpieces fall naturally into the qualified product unloading station as the turntable rotates normally to complete unloading and collection. For unqualified double torsion springs judged to have defects, the controller immediately outputs a sorting control signal, triggering the corresponding pneumatic pushing mechanism to quickly extend and retract, accurately pushing the unqualified workpiece to the unqualified product unloading station, realizing automatic diversion and zoned collection of qualified and unqualified products.

[0048] 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 process, method, article, or apparatus.

[0049] 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 dual-torsion spring imaging detection method, characterized in that, The specific steps are as follows: Step 1: Position the double torsion spring to be tested at the feed position of the turntable; Step Two: The rotation of the turntable drives the double torsion spring through three image acquisition stations in sequence, and collects images of the double torsion spring in different postures. Step 3: Multi-pose images are analyzed and processed to automatically determine whether the molding quality of the double torsion spring is up to standard; Step Four: Based on the judgment results, qualified double torsion springs and unqualified double torsion springs are output separately from different outlets.

2. The dual-torsion spring imaging detection method according to claim 1, characterized in that: The turntable described in step two rotates intermittently in increments, and the double torsion springs precisely pause for a preset duration at each image acquisition station to complete steady-state image acquisition.

3. The dual-torsion spring imaging detection method according to claim 1, characterized in that: The three image acquisition stations mentioned in step two correspond to the frontal, side, and oblique three-dimensional postures of the double torsion spring, respectively, to achieve full-dimensional image acquisition of the double torsion spring.

4. The dual-torsion spring imaging detection method according to claim 1, characterized in that: The image analysis and processing described in step three includes image noise reduction, grayscale processing, contour extraction, standard model comparison, geometric tolerance detection, and automatic identification of double torsion spring dimensional deviations, deformations, burrs, and asymmetric defects.

5. A dual-torsion spring imaging detection device, characterized in that, An imaging detection method for a double torsion spring according to any one of claims 1-4 includes a turntable, wherein the turntable is provided with a feeding position for placing the double torsion spring to be detected; three image acquisition devices for acquiring images of the double torsion spring in different postures are arranged sequentially along the rotation direction of the turntable; and two discharge positions are provided downstream of the three image acquisition devices for outputting qualified and unqualified double torsion springs respectively.

6. The dual-torsion spring imaging detection device according to claim 5, characterized in that: The feed position of the turntable is equipped with a contour positioning fixture that matches the shape of the double torsion spring, which is used to limit and fix the double torsion spring to be tested.

7. The dual-torsion spring imaging detection device according to claim 5, characterized in that: The image acquisition device is equipped with a ring-shaped shadowless supplementary light source, which is coaxially arranged with the image acquisition device.

8. The dual-torsion spring imaging detection device according to claim 5, characterized in that: It also includes an image processing controller, which is electrically connected to three image acquisition devices to receive image data, analyze and determine the quality of workpieces, and output sorting control signals.

9. The dual-torsion spring imaging detection device according to claim 5, characterized in that: Both discharge positions are equipped with pneumatic sorting and pushing mechanisms, which receive signals from the image processing controller to complete precise sorting actions.

Images