Flexible film dynamic following wrinkle removing device based on machine vision
By using machine vision recognition and a dynamic following wrinkle removal device with flexible flattening rollers, the problem of real-time adjustment in traditional film wrinkle removal technology has been solved, achieving high-precision and low-damage film wrinkle removal effect, and adapting to the needs of diverse film materials.
Patent Information
- Application Number
- CN202511825052.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-05
- Publication Date
- 2026-02-13
AI Technical Summary
Existing thin-film wrinkle removal technologies cannot achieve real-time dynamic adjustment of wrinkles, resulting in uneven wrinkle removal effects or damage to the film, especially in high-speed production lines and diverse film materials, making it difficult to meet the requirements for precise processing.
A machine vision-based flexible film dynamic tracking wrinkle removal device is adopted. The device detects the wrinkle position in real time through a vision recognition module, and combines the precise positioning control of the flexible flattening roller and the thrust actuator to achieve dynamic tracking and localized processing of film wrinkles.
It improves wrinkle removal accuracy and efficiency, reduces the risk of damage to film materials, adapts to rapid changes in wrinkle position and shape in high-speed production lines, and provides personalized wrinkle removal treatment for film materials of different types and thicknesses.
Smart Images

Figure CN121516629A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of flexible film processing and intelligent manufacturing, and particularly relates to a flexible film dynamic following wrinkle removing device based on machine vision. BACKGROUND
[0002] Flexible film materials are widely used in modern industrial production, including electronic device manufacturing, packaging material processing, optical film production, etc. In the process of film transmission and processing, due to factors such as material properties, transmission tension changes, equipment vibration, etc., wrinkles are easily generated on the surface of the film. Traditional film wrinkle removing techniques mainly rely on mechanical fixed structures, such as flattening rollers, tension adjusting devices, etc., which remove wrinkles by applying constant mechanical force. These methods have been widely used in film processing equipment and have played a role in maintaining film flatness and processing quality.
[0003] The existing wrinkle removing techniques have some limitations. Mechanical fixed structures cannot be adjusted in real time according to the actual position and degree of wrinkles, and often use constant processing methods for the full width, which may result in uneven wrinkle removing effect or cause undue stress to the film. For high-speed running film production lines, the position and shape of wrinkles change quickly, and traditional static wrinkle removing methods are difficult to achieve accurate dynamic response. In addition, different types and thicknesses of film materials have different requirements for wrinkle removing force, and fixed parameter wrinkle removing devices are difficult to adapt to diversified production needs, which may cause material damage when processing high rigidity films. SUMMARY
[0004] In view of the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide a flexible film dynamic following wrinkle removing device based on machine vision, which can improve the wrinkle removing precision and efficiency, and reduce the risk of damage to the film material by real-time detection of wrinkle position and dynamic adjustment of wrinkle removing parameters through machine vision.
[0005] To achieve the above-mentioned purpose, the present application provides a flexible film dynamic following wrinkle removing device based on machine vision, which adopts the following technical solutions: A vision recognition module for real-time image acquisition and wrinkle position recognition of the running flexible film; A flexible flattening roller made of flexible material for contacting the flexible film; A thrust actuator arranged on the back of the flexible flattening roller, including a thrust component that can move axially along the flexible flattening roller; A moving device for driving the thrust component to move axially to the wrinkle position; and A control unit is configured to control the thrust actuator according to the wrinkle position identified by the vision recognition module, so that the thrust component is aligned with the wrinkle position and pushes the flexible flattening roller to generate local deformation, thereby forcibly flattening the film wrinkle.
[0006] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the vision recognition module comprises a high-speed linear array camera configured to collect linear array images of the running flexible film in real time.
[0007] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the control unit is internally provided with an image processing algorithm configured to perform real-time segmentation and analysis on the linear array images to identify the position and width of the wrinkle.
[0008] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the image processing algorithm comprises preprocessing the collected linear array images, including noise filtering, contrast enhancement, and image correction; identifying the gray scale change and texture features of the film surface through an edge detection algorithm; determining the geometric center of the wrinkle area through centroid calculation, and converting the pixel position of the wrinkle in the image coordinate system into the physical position in the actual width direction of the film; analyzing the boundary contour of the wrinkle area through a contour analysis method, and calculating the effective width range of the wrinkle; based on the multi-dimensional feature parameters of the geometric shape, area size, length-width ratio, and gray scale distribution of the wrinkle area, judging whether the detected area is a real wrinkle through a pre-set feature template; outputting the accurate position and width information of the wrinkle, and providing accurate parameter basis for the precise positioning of the thrust actuator and the local deformation control of the flexible flattening roller.
[0009] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the flexible flattening roller is made of soft rubber material and has bendability.
[0010] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the control system of the driving motor comprises a position controller configured to receive target position instructions from the control unit, acquire real-time position information of the driving motor through encoder feedback, calculate position deviation, and output speed instructions; a speed regulator configured to adjust the running speed of the motor through a PID algorithm according to the speed instructions of the position controller and motor speed feedback signals, and ensure that the thrust component moves to the target position at an appropriate speed; A current driver is configured to receive the output signal of the speed regulator, control the magnitude and direction of the motor winding current, and generate corresponding torque to drive the motor to rotate; A protection circuit is configured to monitor the working state of the motor and provide overload protection and fault alarm in abnormal conditions; A communication interface is configured to realize data exchange and instruction transmission between the control system and the upper control unit, so as to ensure that the driving motor can accurately respond to the wrinkle detection result and realize accurate positioning control of the thrust component when the wrinkle position instruction is received.
[0011] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the moving device comprises: An upper slide rail and a lower slide rail are configured to guide the axial movement of the thrust component; and A driving motor is configured to drive the thrust component to move along the upper slide rail and the lower slide rail.
[0012] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the thrust execution mechanism further comprises: A universal ball is configured to adapt to the arc-shaped deformation of the flexible flattening roller; and A pushing cylinder is configured to contact the flexible flattening roller through the universal ball and apply a pushing force.
[0013] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, it further comprises a front wrinkle vision recognition module and a back wrinkle vision recognition module, which are respectively used for detecting the wrinkles on the front and back of the flexible film.
[0014] Further, in the above-mentioned machine vision-based flexible film dynamic following wrinkle removal device, the control unit converts the pixel coordinates of the wrinkles into mechanical coordinates to realize real-time positioning control of the thrust component.
[0015] In summary, compared with the prior art, the present application has at least one of the following beneficial technical effects: The machine vision-based flexible film dynamic following wrinkle removal device of the present application can realize dynamic following and localized processing of film wrinkles by real-time detection of wrinkle positions by the vision recognition module, combined with accurate positioning control of the flexible flattening roller and the thrust execution mechanism. Compared with the traditional full-width constant wrinkle removal method, the present application can significantly improve the wrinkle removal accuracy and efficiency, reduce unnecessary stress and damage risk to the film material, and adapt to the rapid changes of wrinkle positions and shapes in high-speed production lines, thereby providing personalized wrinkle removal processing solutions for different types and thicknesses of film materials. BRIEF DESCRIPTION OF DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.
[0017] Figure 1 A schematic diagram of a thin film wrinkle removal and flattening device in the prior art is shown.
[0018] Figure 2 A general structural diagram of a flexible thin film dynamic following wrinkle removal device based on machine vision is shown.
[0019] Figure 3 A flowchart of a flexible thin film wrinkle detection image processing algorithm is shown.
[0020] Figure 4 A detailed structural diagram of a thrust actuator and a moving device is shown.
[0021] Figure 5 A flowchart of the control unit closed-loop control logic is shown. DETAILED DESCRIPTION
[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application. In addition, it should be understood that the specific embodiments described herein are only used to illustrate and explain the present application, and are not used to limit the present application.
[0023] It should be noted that the description order of the following embodiments is not used to limit the preferred order of the embodiments of the present application. In the following embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.
[0024] The execution order of the method steps described in the embodiments of the present application can be executed in the order described in the specific embodiments, or the execution order of each step can be adjusted on the premise of solving the technical problems according to actual needs, which is not listed one by one here.
[0025] The present application will be further described in detail below with reference to the drawings.
[0026] Reference Figure 1 The existing thin film wrinkle removal technology mainly uses mechanical fixed structure to process the wrinkle problem of flexible thin film generated in the conveying process. Figure 1A conventional film wrinkle flattening device is shown, which includes a mechanical system composed of multiple tension rollers and flattening rollers. After the film enters the system from the unwinding end, it passes through the first tension roller 11, the second tension roller 12 and the third tension roller 13 in turn. These tension rollers form a serpentine transmission path by different height configurations, and exert a constant tension on the film.
[0027] In some embodiments, the conventional wrinkle removal device is also configured with a flattening roller structure, including a first flattening roller 21 and a second flattening roller 22. The first flattening roller 21 is composed of an upper roller 211 and a lower roller 212, which mechanically flattens the film through the pressing action of the upper and lower rollers. This fixed mechanical structure eliminates the wrinkles on the surface of the film through continuous contact pressure and tension adjustment.
[0028] However, such traditional methods have significant technical limitations. The existing mechanical fixed structure cannot respond and dynamically adjust to the actual occurrence position of the wrinkles in real time. During high-speed conveying of the film, even under tension, wrinkles may occur at different positions due to slight vibration or changes in material properties, and the traditional flattening roller can only exert constant force at a fixed position, and cannot achieve precise processing of specific wrinkle areas.
[0029] In some embodiments, for thin film materials with high rigidity, the conventional wrinkle removal device needs to exert greater tension to achieve flattening, but high tension often leads to film tearing or permanent deformation, causing material damage. In addition, the existing technology lacks the ability to detect and identify the position of the wrinkles, and cannot achieve responsive wrinkle removal, often resulting in incomplete wrinkle removal or unnecessary damage to the film.
[0030] These technical defects of the conventional wrinkle removal device limit its effectiveness in high-precision thin film processing applications, especially in situations where the integrity and surface quality of the film need to be maintained. The existing technology cannot meet the technical requirements of dynamic controllable processing and precise wrinkle removal.
[0031] Reference Figure 2 The present application provides a flexible film dynamic following wrinkle removal device based on machine vision, which realizes intelligent wrinkle removal of running flexible film by integrating multiple functional modules. The overall system structure of the device includes a vision recognition module, a flexible flattening roller, a thrust execution mechanism, a moving device and a control unit, and other main components. These components cooperate in the system to achieve accurate detection and dynamic wrinkle removal of the film.
[0032] In some embodiments, the device is configured with front and rear over rollers 21 and 22 to guide the transmission path of the flexible film 5 within the wrinkle removal area. The film 5 enters the system from the gap between the front and back wrinkle visual recognition modules 11 and 12, and exits the system through the rear over roller 22 after passing through the processing area of the flexible flattening roller. The front and rear over rollers 21 and 22 provide stable guiding support for film transmission, ensuring that the film maintains proper tension and transmission speed during processing.
[0033] Referring to Figure 2 , the visual recognition module includes a front wrinkle visual recognition module 11 and a back wrinkle visual recognition module 12, which are respectively arranged above and below the film transmission path. The front wrinkle visual recognition module 11 is used to collect real-time images of the running flexible film and identify the wrinkle position on the front of the film, and the back wrinkle visual recognition module 12 is used to detect possible wrinkles on the back of the film. This double-sided detection configuration enables the system to comprehensively monitor the surface state of the film, improving the accuracy and completeness of wrinkle detection.
[0034] In some embodiments, the flexible flattening roller is made of flexible material for contact with the flexible film. The flexible flattening roller is located at the center of the system, and the film 5 contacts the surface of the flexible flattening roller during transmission. The thrust actuator 4 is arranged on the back of the flexible flattening roller, and the thrust actuator 4 includes a thrust component that can move axially along the flexible flattening roller.
[0035] Referring to Figure 2 , the moving device 41 is used to drive the thrust component to move axially to the wrinkle position. The moving device 41 controls the precise positioning of the thrust component to accurately move to the wrinkle position detected by the visual recognition module. The thrust actuator 4 is driven by the moving device 41 to achieve precise positioning and movement along the axis of the flexible flattening roller.
[0036] In some embodiments, the system further includes side plates 6 as a structural support assembly to provide mechanical stability and precise component positioning for the entire device. The side plates 6 fix and support various functional modules to ensure that the system maintains a stable geometric relationship and precise relative position during operation.
[0037] The control unit is used to control the thrust actuator according to the wrinkle position identified by the visual recognition module, so that the thrust component is aligned with the wrinkle position and pushes the flexible flattening roller to produce local deformation, thereby forcibly flattening the film wrinkles. The system realizes the complete processing flow of visual recognition → coordinate conversion → target positioning → local pushing and flattening through a closed-loop control logic sequence.
[0038] In some embodiments, the thrust mechanism is driven by a moving device to align and track the moving wrinkles on the film, achieving a dynamic follow-up forced flattening function. When the visual recognition module detects the position of the wrinkles, the control unit calculates the target position that the thrust component needs to move to, and the moving device then drives the thrust component to move to the specified position. The thrust component applies local thrust to the flexible flattening roller, causing controllable local deformation of the flexible flattening roller at the position corresponding to the wrinkles, thereby exerting flattening force on the film wrinkles.
[0039] In some embodiments, the visual recognition module includes a high-speed line array camera specially configured for real-time line array image acquisition of the running flexible film. The high-speed line array camera uses line array sensor technology and can perform line-by-line scanning and acquisition of the continuously moving film surface at a high frequency, forming a continuous line array image data stream. The scanning frequency of the line array camera is synchronized and matched with the running speed of the film, ensuring that the acquired images have sufficient resolution and clarity to detect the wrinkle features on the film surface.
[0040] The high-speed line array camera performs transverse scanning along the width direction of the film through its linear sensor array, and acquires image information of the film surface line by line as the film moves longitudinally. This line array acquisition method is particularly suitable for processing continuously moving film materials. Compared with traditional area array cameras, line array cameras can achieve higher acquisition speed and more precise synchronization control, avoiding image blur or distortion caused by film movement.
[0041] In some embodiments, the control unit is built-in with an image processing algorithm specially used for real-time segmentation and analysis of line array images to identify the position and width of wrinkles. The image processing algorithm receives continuous image data from the high-speed line array camera, and through digital image processing techniques, it pre-processes, extracts features and recognizes patterns of the images, thereby accurately identifying the wrinkle areas on the film surface.
[0042] The image processing algorithm first performs pre-processing operations on the acquired line array images, including noise filtering, contrast enhancement and image correction, etc., to improve image quality and accuracy of subsequent analysis. The pre-processed images are identified by edge detection algorithms for gray scale changes and texture features on the film surface. The wrinkle areas will exhibit different gray scale distribution and edge features in the image due to their special optical properties, which are different from the flat areas.
[0043] In some embodiments, with reference to Figure 3, the control unit adopts a multi-stage image processing procedure based on parameterized configuration to ensure high-precision detection of the wrinkles even in the scenario of high-speed film running. Specifically, first, a pre-processing operation is performed on the original image output by the high-speed linear array camera, wherein a median filter with a window size of 5x1 can be used for noise filtering, so that the high-frequency noise is reduced by not less than 30%, and meanwhile, the overall contrast of the image is improved by 15%-25% through adaptive histogram equalization, so as to enhance the gray scale demarcation of the wrinkle region.
[0044] In the pre-processed image, the Canny edge detection operator (low threshold 40, high threshold 120) is further used to extract the film surface region with rapid gray scale change. Experiments show that the real wrinkles usually have a gradient intensity of not less than 0.45 (based on the normalized gradient amplitude) in the edge map, while the gradient of the flat region is generally less than 0.25, so the gradient intensity threshold T1 = 0.35 is set for the preliminary screening of the candidate wrinkle region.
[0045] For the screened candidate region, the centroid of the pixel coordinates in the region is calculated. The centroid (Xc, Yc) adopts the standard formula: Xc = Σ(xi·wi) / Σ wi; Yc = Σ(yi·wi) / Σ wi; wherein the pixel weight wi is the pixel gray scale difference |Ii - I|. The calculated pixel coordinates are converted into physical coordinates according to the calibration parameters. For example, in the present embodiment, the pixel pitch of the linear array camera is 14 μm, and the scaling coefficient of the actual width direction of the film is between 1.000 and 1.004, so the pixel position can be converted into a physical transverse coordinate with a precision of ±0.2 mm through the correction matrix.
[0046] After determining the center position of the wrinkle, the profile analysis is performed on the candidate region. The 8-neighborhood tracking algorithm is used to extract the upper and lower boundaries to measure the effective width W of the wrinkle. The slight wrinkle generally has a width of 0.6 mm≤W≤1.5 mm, the moderate wrinkle has a width of 1.5 mm~3 mm, and the severe wrinkle has a width of more than 3 mm. In the present embodiment, the "effective width threshold A1" is set to W≥0.5 mm to exclude the tiny noise region.
[0047] In some embodiments, a preset feature template is used for multi-dimensional matching judgment of the candidate region, which includes but is not limited to the following specific parameter conditions: gray contrast feature, the average gray scale of the wrinkle center region needs to be at least ΔG1 = 18 (calculated in the 0-255 gray scale range) lower than the average gray scale of the adjacent flat region; gradient feature, the average gradient intensity of the wrinkle boundary needs to be ≥T1 = 0.35, and the maximum gradient needs to be ≥0.55; Area feature, the projected area of the wrinkle needs to be ≥ 50 pixels 2 Corresponding to the physical area of the film is not less than 0.01 mm 2 ); Aspect ratio feature, the aspect ratio of the wrinkle area needs to meet L / W ≥ R1 = 1.5; Direction consistency, the change range of the main direction vector of the wrinkle needs to be within ±θ1 = 12°, which is used to exclude noise textures; Gray standard deviation, the gray standard deviation of the wrinkle area needs to be ≥ σ1 = 8, so as to exclude low-texture background areas.
[0048] If the candidate area meets the threshold requirements of most (for example, at least 4 of the 6 conditions), then a comprehensive evaluation is made based on a confidence score (0-1), and when the confidence is ≥ 0.65, it is determined to be a real wrinkle, otherwise it is considered to be an artifact or a film surface texture.
[0049] The final confirmed wrinkle position and width parameters are output to the control unit for driving the thrust actuator to achieve local precise deformation, thereby completing the real-time dynamic following and flattening of the film wrinkle.
[0050] In some embodiments, the image processing algorithm calculates the accurate position and width information of the identified wrinkle. The position calculation is achieved by determining the centroid coordinates or geometric center of the wrinkle area, and the algorithm converts the pixel position of the wrinkle in the image coordinate system to the actual physical position relative to the film width. The width calculation is determined by measuring the span of the wrinkle area in the transverse direction of the film, and the algorithm analyzes the boundary profile of the wrinkle area to calculate the effective width range of the wrinkle, providing accurate geometric parameters for subsequent thrust component positioning and deformation control.
[0051] In some embodiments, the machine vision-based flexible film dynamic following and wrinkle removal device also includes a front wrinkle visual recognition module and a back wrinkle visual recognition module, respectively used for detecting the front and back wrinkles of the flexible film. This double-sided detection configuration realizes synchronous monitoring and wrinkle detection of the two surfaces of the film by setting independent visual recognition modules above and below the film transmission path.
[0052] The front wrinkle visual recognition module is arranged at a position above the film transmission path, and its optical axis is vertically directed to the upper surface of the film. The module continuously collects images of the front of the film through a high-speed line array camera, and during the collection process, the scanning line of the camera maintains a perpendicular relationship with the width direction of the film, ensuring that it can completely cover the entire width range of the front of the film. The front wrinkle visual recognition module is configured with a special lighting system, which irradiates the front of the film with uniform light sources to provide stable lighting conditions for image collection.
[0053] The back-side crease visual recognition module is arranged at a position below the film transmission path, with its optical axis pointing upwards to the lower surface of the film. The module uses the same line array camera technology as the front-side crease visual recognition module to capture real-time images of the back surface of the film. The illumination system of the back-side crease visual recognition module shines from below to the back surface of the film, with the angle and intensity of illumination optimized to ensure that the crease features on the back surface are clearly visible in the images.
[0054] In some embodiments, the front-side crease visual recognition module and the back-side crease visual recognition module use a synchronous acquisition mechanism, with the image acquisition timing of the two modules kept in precise synchronization to ensure simultaneous detection of the front and back surfaces of the film at the same cross-sectional position. Synchronous acquisition is achieved through unified clock signal control, with both camera modules receiving the same trigger signal and starting image acquisition simultaneously during film movement to obtain corresponding image data of the front and back surfaces of the film.
[0055] The double-sided detection configuration significantly improves the comprehensiveness of crease recognition. Creases may occur on either surface of the film during transmission, and single-sided detection may miss crease problems that only occur on the other surface. Through the cooperative work of the front-side crease visual recognition module and the back-side crease visual recognition module, the system can comprehensively monitor the overall surface state of the film, ensuring that any creases that occur at any position are detected in a timely manner.
[0056] In some embodiments, double-sided detection also improves the accuracy of crease recognition. Some creases may not be clearly visible on one surface of the film, but have clearer characteristic manifestations on the other surface. By comparing and analyzing the image data of the front and back surfaces, the system can cross-verify the existence and position of the creases, reducing the incidence of false positives and false negatives. Fusion analysis of double-sided detection data can also provide three-dimensional feature information of the creases, including the depth, direction, and morphological characteristics of the creases, providing more accurate references for subsequent thrust control.
[0057] The detection results of the front-side crease visual recognition module and the back-side crease visual recognition module are comprehensively processed through a data fusion algorithm. The data fusion algorithm performs spatial registration and feature matching on the image data from the two modules, integrating the crease information detected by the front and back surfaces into a unified crease distribution map. When the front and back surfaces simultaneously detect a crease at the same position, the system confirms the existence of the crease at that position and increases the confidence; when only one surface detects a crease, the system verifies the reliability of the detection result through neighborhood analysis and historical data comparison.
[0058] In some embodiments, the double-sided detection system is also equipped with a wrinkle type identification function. By analyzing the different performance characteristics of the wrinkle on the front and back surfaces, the system can distinguish between different types of wrinkles, such as surface wrinkles, penetrating wrinkles, or deformations caused by internal stress of the material. This wrinkle type identification capability enables the system to adopt corresponding wrinkle removal strategies for different types of wrinkles, optimize the thrust control parameters, and improve the effectiveness of wrinkle removal.
[0059] In some embodiments, the flexible flattening roller is made of soft rubber material and has bendability. The selection of soft rubber material provides the flexible flattening roller with appropriate elastic modulus and deformation capability, enabling it to produce controllable shape changes under external force without permanent deformation or structural damage. Soft rubber material has good elastic recovery characteristics, and the flexible flattening roller can quickly recover to its original cylindrical shape after the external force applied by the thrust component is removed.
[0060] The bendability of soft rubber material enables the flexible flattening roller to produce local geometric deformation under the action of the thrust component. This bendability is derived from the molecular structure characteristics of the rubber material, and the rubber molecular chain can undergo reversible conformational changes under external force, enabling the entire roller body to exhibit elastic deformation capability in a macroscopic manner. The wall thickness and hardness parameters of the flexible flattening roller are optimized to ensure appropriate deformation response characteristics within the normal working load range.
[0061] In some embodiments, the flexible flattening roller produces inward or outward arc bending through the alignment and force application of the thrust component. When the thrust component applies a thrust force from the back surface of the flexible flattening roller, the roller body material around the thrust action point undergoes local compression and deformation, resulting in a convex deformation of the flexible flattening roller in the direction of the film at that area, forming an outward arc bending. This outward arc bending causes the flexible flattening roller to form a local convex profile at the position corresponding to the wrinkle, increasing the contact pressure between that area and the film.
[0062] When the action mode or installation configuration of the thrust component is adjusted, the flexible flattening roller can also produce inward arc bending. Inward arc bending changes the curvature radius of the local concave area formed on the surface of the flexible flattening roller, thereby producing different tension distribution effects on the film passing through that area. Inward and outward arc bending provides the system with diversified wrinkle removal processing modes, which can adapt to different types and severity of film wrinkles.
[0063] The flexible flattening roller produces controllable arc deformation, and the degree and position of deformation are accurately adjusted through precise control of the thrust mechanism. The force applied by the thrust component directly affects the degree of deformation of the flexible flattening roller. A small thrust produces slight local deformation, which is suitable for handling slight film wrinkles. A larger thrust produces more obvious arc deformation, which is used to handle more serious wrinkle problems. The controllability of deformation is realized through the force adjustment system of the thrust component, which can automatically adjust the thrust size according to the wrinkle detection results.
[0064] In some embodiments, the flexible flattening roller realizes non-contact local wrinkle removal by applying stretching force to the film wrinkle through local deformation. When the flexible flattening roller produces arc deformation at the wrinkle position, the curvature change of the deformation area changes the stress distribution state of the film in this area. The arc deformation makes the film in the wrinkle area subjected to radial and tangential combined stretching, and the radial stretching force unfolds the wrinkle outward, and the tangential stretching force eliminates the longitudinal contraction of the wrinkle.
[0065] The working principle of non-contact local wrinkle removal is based on the change of film geometric constraint caused by the deformation of the flexible flattening roller. The local arc deformation of the flexible flattening roller creates a dynamically changing support surface, and the film is forced to adapt to the new geometric profile when passing through the deformation area. The stretching effect caused by this change in geometric constraint can effectively eliminate wrinkles without additional mechanical contact or pressure application, avoiding damage or indentation to the film surface that may be caused by traditional wrinkle removal methods.
[0066] In some embodiments, the stretching force distribution generated by the local deformation of the flexible flattening roller has spatial selectivity, and the stretching effect is mainly concentrated in the wrinkle area and its adjacent range, and has little effect on other areas of the film. This localized force distribution characteristic enables the system to achieve precise wrinkle removal treatment and avoid unnecessary disturbance to the overall tension state of the film. The size and distribution range of local stretching force are adjusted by the deformation degree and deformation profile of the flexible flattening roller, providing adaptive wrinkle removal treatment parameters for different specifications and characteristics of film materials.
[0067] Reference Figure 4 The mobile device includes an upper slide rail 411 and a lower slide rail 412 for guiding the axial movement of the thrust component. The upper slide rail 411 and the lower slide rail 412 constitute a parallel guide rail system extending in the axial direction of the flexible flattening roller, providing precise linear motion guidance for the thrust component. The upper slide rail 411 is arranged at the upper position of the thrust component, and the lower slide rail 412 is arranged at the lower position of the thrust component. The two slide rails maintain a fixed distance and parallel relationship, ensuring that the thrust component maintains a stable posture and accurate motion trajectory during movement.
[0068] In some embodiments, the upper slide rail 411 and the lower slide rail 412 adopt a precision machined metal guide rail structure with high precision surface finish and geometric accuracy. The slide rail surface is specially treated to reduce frictional resistance, ensuring that the thrust component can move smoothly in the axial direction. The upper slide rail 411 and the lower slide rail 412 are fixed on the chassis of the device through a support structure, which provides sufficient rigidity to withstand the load and vibration generated during the movement of the thrust component.
[0069] The thrust component is connected to the upper slide rail 411 and the lower slide rail 412 through a sliding block mechanism, which includes a guide sliding block matched with the slide rail and a bearing assembly. The guide sliding block slides along the channel or guide surface of the slide rail, and the bearing assembly reduces sliding friction and improves movement accuracy. The thrust component can only move in the axial direction under the constraint of the upper slide rail 411 and the lower slide rail 412, and displacement and rotation in other directions are effectively limited, ensuring that the movement trajectory of the thrust component remains parallel to the axis of the flexible flattening roller.
[0070] Referring to Figure 4 , the mobile device further includes a drive motor 413 for driving the thrust component to move along the upper slide rail 411 and the lower slide rail 412. The drive motor 413 is arranged at one end of the slide rail system and connected to the thrust component through a transmission mechanism, providing a power source for the axial movement of the thrust component. The drive motor 413 adopts a servo motor or a stepper motor, which has precise position control capability and fast response characteristics, and can realize precise positioning of the thrust component according to the instructions of the control unit.
[0071] In some embodiments, the drive motor 413 drives the thrust component to move through a lead screw transmission mechanism. The lead screw transmission mechanism includes a lead screw connected to the drive motor 413 and a nut connected to the thrust component, and the rotational movement of the drive motor 413 is converted into linear movement of the thrust component through the lead screw-nut pair. The lead screw transmission mechanism has high precision position transmission characteristics and good self-locking ability, ensuring that the thrust component can accurately move to the specified position and remain stable at that position.
[0072] The control system of the drive motor 413 receives position instructions from the control unit, calculates the distance and direction that the thrust component needs to move according to the wrinkle position detected by the visual recognition module. The control system obtains real-time position information of the thrust component through encoder feedback, forming a closed-loop position control, ensuring that the thrust component can accurately move to the axial position corresponding to the wrinkle. The motion speed and acceleration parameters of the drive motor 413 are optimized according to the running speed of the film and the movement characteristics of the wrinkle, realizing effective tracking of dynamic wrinkles.
[0073] Referring to Figure 4The thrust actuator further comprises a universal ball 421 for adapting to the arc deformation of the flexible flattening roller. The universal ball 421 is arranged at the front end of the thrust component as a contact element for thrust transmission and contacts the back surface of the flexible flattening roller. The universal ball 421 has a spherical contact surface and can freely rotate and adjust the contact angle in multiple directions to adapt to various arc deformation states of the flexible flattening roller under the action of thrust.
[0074] In some embodiments, the spherical design of the universal ball 421 enables it to maintain a good contact state with the curved surface of the flexible flattening roller. When the flexible flattening roller is deformed in an arc shape under the action of thrust, the surface curvature of the deformed area changes, and the universal ball 421 automatically adjusts the contact angle through its spherical geometric properties to ensure that the contact surface always maintains appropriate contact with the surface of the flexible flattening roller. This self-adaptive contact capability avoids stress concentration or poor contact problems caused by improper contact angle.
[0075] The material selection of the universal ball 421 considers the requirements of contact strength and wear resistance, and high-strength metal materials or engineering plastics are used. The surface of the universal ball 421 is treated to be smooth to reduce frictional resistance when contacting the flexible flattening roller, avoiding wear or damage to the surface of the flexible flattening roller. The size and curvature radius of the universal ball 421 are optimized according to the diameter of the flexible flattening roller and the expected degree of deformation to ensure effective thrust transmission under various working conditions.
[0076] Referring to Figure 4 The thrust actuator further comprises a push cylinder 423 for contacting the flexible flattening roller through the universal ball 421 and applying thrust. The push cylinder 423 serves as a thrust generating device and generates controllable thrust output through the extension and retraction movement of the piston rod driven by gas pressure. The piston rod of the push cylinder 423 is connected with the universal ball 421 to transmit the thrust generated by the cylinder to the back surface of the flexible flattening roller through the universal ball 421.
[0077] In some embodiments, the push cylinder 423 adopts a precise pneumatic cylinder structure with good force control precision and response speed. The gas pressure supply system of the push cylinder 423 includes a pressure regulating valve and a flow control valve to control the output force and action speed of the push cylinder 423 by adjusting the gas supply pressure and flow parameters. The thrust size of the push cylinder 423 is adjusted according to the severity of the wrinkles and the characteristics of the film material to ensure that appropriate deformation force is applied to achieve effective wrinkle removal without damaging the film.
[0078] The push cylinder 423 controls the contact state of the universal ball 421 with the flexible flattening roller through its telescopic action. When the wrinkle needs to be removed, the push cylinder 423 is extended to push the universal ball 421 to contact the back of the flexible flattening roller and apply a pushing force, so that the flexible flattening roller is locally deformed at this position. When the wrinkle removal is completed or needs to be moved to other positions, the push cylinder 423 is retracted, the universal ball 421 is separated from the flexible flattening roller, and the flexible flattening roller returns to its original shape.
[0079] Referring to Figure 4 The thrust actuator further comprises a connector 422 for connecting the universal ball 421 and the push cylinder 423. The connector 422 is arranged between the universal ball 421 and the piston rod of the push cylinder 423 to realize the mechanical connection and thrust transmission of the two components. The design of the connector 422 takes into account the multi-directional rotation requirement of the universal ball 421 and the linear motion characteristics of the push cylinder 423, and realizes the motion coordination between the two through appropriate connection structure.
[0080] In some embodiments, the connector 422 adopts a universal joint or spherical joint structure, allowing the universal ball 421 to rotate and swing within a certain angle range relative to the piston rod of the push cylinder 423. This flexible connection design enables the universal ball 421 to adapt to the angle change of the flexible flattening roller when deformed, while maintaining effective thrust transmission between the push cylinder 423. The connector 422 has sufficient strength to withstand the maximum output force of the push cylinder 423, ensuring reliable thrust transmission under various working loads.
[0081] The structure of the connector 422 also takes into account the convenience of assembly and maintenance, realizing the connection with the universal ball 421 and the push cylinder 423 through threaded connection or quick connector. The material selection of the connector 422 takes into account the strength and corrosion resistance requirements, and is made of stainless steel or alloy steel material, ensuring good mechanical properties and connection reliability during long-term use.
[0082] Referring to Figure 5 The control unit realizes intelligent detection and dynamic wrinkle removal of the flexible film wrinkle through a closed-loop control logic process. This control logic process establishes a complete automation sequence from visual recognition to mechanical execution, realizing accurate response and processing of the film wrinkle through the coordination of multiple functional modules. The control logic process starts from the start node, establishes the initialization state and parameter configuration of the system, and provides the basis for subsequent detection and processing operations.
[0083] In some embodiments, the first processing step of the control logic flow is the image acquisition stage. The image acquisition module initiates the high-speed line array camera to continuously capture real-time images of the flexible film passing through the detection area, ensuring clear and complete image data of the film surface. The image acquisition stage establishes the data input basis for the entire control flow, providing raw image information for subsequent wrinkle detection and analysis.
[0084] After image acquisition is complete, the control logic flow enters the wrinkle detection judgment stage. This stage analyzes the collected image data through image processing algorithms to determine whether there are wrinkle characteristics on the film surface. The wrinkle detection uses preset judgment conditions, including gray scale change threshold, edge feature intensity, and area size, to determine the presence of wrinkles by analyzing these characteristic parameters.
[0085] In some embodiments, when the wrinkle detection result is "none", the control logic flow returns to the image acquisition stage, forming a continuous monitoring loop. This loop monitoring mechanism ensures that the system can continuously track changes in the film surface state and timely detect new wrinkles. The frequency of loop monitoring is consistent with the frame rate of image acquisition, achieving real-time monitoring coverage of the film surface.
[0086] When the wrinkle detection result is "yes", the control logic flow enters the AI positioning stage. The AI positioning stage uses advanced image processing algorithms to accurately calculate the position and analyze the characteristics of the detected wrinkles, determining the specific position, width range, and geometric characteristics of the wrinkles in the image coordinate system. The AI positioning algorithm uses pixel-level analysis accuracy, using methods such as centroid calculation, boundary detection, and contour analysis to obtain accurate geometric parameters of the wrinkles.
[0087] Referring to Figure 5 After the AI positioning stage is complete, the control logic flow enters the coordinate conversion stage. The coordinate conversion stage is a key link for the control unit to achieve accurate positioning control, which converts the pixel coordinates of the wrinkles in the image coordinate system to physical position coordinates in the mechanical coordinate system. The coordinate conversion process establishes an accurate correspondence between the visual detection results and the mechanical execution actions, providing a coordinate reference for accurate positioning of the thrust component.
[0088] In some embodiments, the control unit uses a pre-established coordinate conversion matrix to perform conversion calculations from pixel coordinates to mechanical coordinates. The coordinate conversion matrix contains the geometric transformation relationship between the image coordinate system and the mechanical coordinate system, and is obtained through a system calibration process. In the calibration process, standard markers are placed at known mechanical positions, and the marker images are captured by the vision system to establish a correspondence between pixel positions and actual physical positions.
[0089] The coordinate conversion calculation considers multiple geometric factors, including the installation angle of the camera, the optical distortion of the lens, the distance change between the film and the camera, etc. The control unit eliminates the influence of lens optical distortion on coordinate accuracy through a distortion correction algorithm, and corrects the image geometric deformation caused by the camera installation angle through perspective transformation, ensuring that the converted mechanical coordinates have high precision and reliability.
[0090] In some embodiments, the control unit also considers the dynamic factors of film movement during coordinate conversion. Due to the time delay between detection and processing, the actual position of the wrinkle will change with the movement of the film. The control unit calculates the expected position of the wrinkle at the processing time through a motion compensation algorithm, which is based on parameters such as the running speed of the film, the detection delay time, and the response time of the thrust component, to make a forward-looking prediction and compensation of the wrinkle position.
[0091] After the coordinate conversion is completed, the control logic flow enters the push positioning phase. The push positioning phase controls the thrust component to move to the axial position corresponding to the wrinkle according to the coordinate conversion result, realizing the accurate alignment of the thrust component and the wrinkle position. Push positioning is achieved through precise control of the drive motor, and the control unit sends position instructions to the drive motor, which drives the thrust component to move to the target position along the slide rail system according to the instructions.
[0092] In some embodiments, the push positioning phase adopts a closed-loop position control strategy to ensure positioning accuracy. The actual position of the thrust component is monitored in real time by a position sensor or encoder, and the control unit compares the deviation between the actual position and the target position, and adjusts the motion parameters of the drive motor through a feedback control algorithm to realize precise positioning of the thrust component. The closed-loop control strategy can effectively compensate for the influence of mechanical transmission errors, load changes, and external disturbances on positioning accuracy.
[0093] After the push positioning is completed, the control logic flow enters the deformation wrinkle removal phase. The deformation wrinkle removal phase controls the thrust component to exert a pushing force on the flexible flattening roller through the action of the push cylinder, so that the flexible flattening roller produces local arc deformation at the position corresponding to the wrinkle, thereby implementing forced flattening processing of the film wrinkle. The pushing force and action time of the deformation wrinkle removal phase are adjusted according to the severity of the wrinkle and the material properties of the film.
[0094] Reference Figure 5 After the deformation wrinkle removal phase is completed, the control logic flow returns to the image acquisition phase, forming a complete closed-loop control cycle. This closed-loop control mechanism enables the system to continuously monitor the surface state of the film, timely discover and process new wrinkle problems, and realize continuous tracking and processing of dynamic wrinkles. The cycle frequency of the closed-loop control is optimized according to the running speed of the film and the frequency of wrinkle occurrence, ensuring that the system has sufficient response speed and processing capacity.
[0095] In some embodiments, the closed-loop control logic of the control unit also includes abnormality handling and fault diagnosis functions. When the system detects abnormal situations such as image acquisition failure, thrust component positioning timeout, or abnormal cylinder action, etc., the control logic initiates corresponding abnormality handling procedures, including retry operations, alarm prompts, and safety shutdown measures, to ensure the reliability and safety of system operation.
[0096] The control unit achieves precise tracking and handling of dynamic wrinkles by real-time positioning control. Real-time positioning control is based on continuous visual feedback and fast coordinate transformation calculations, enabling the thrust component to dynamically adjust its axial position and always align with the moving wrinkle area on the film. This real-time tracking capability breaks through the limitations of traditional static wrinkle removal methods, effectively handling transient and moving wrinkles, significantly improving the adaptability and processing effect of the wrinkle removal system.
[0097] In some embodiments, the control system of the driving motor includes a position controller, a speed regulator, a current driver, a protection circuit, and a communication interface. The position controller is configured to receive a target position instruction from the control unit and acquire the current position of the motor in real time through an encoder installed at the shaft end of the driving motor. The encoder can be an incremental encoder with a pulse per revolution (ppr) of 1000-2500, and the sampling frequency can be 2 kHz-5 kHz to ensure that the position acquisition accuracy is not less than ±0.05 mm. The position controller compares the real-time position information with the target position, calculates the position deviation, and generates a speed instruction according to the control parameters of the proportional gain Kp=1.2-2.0 and the differential gain Kd=0.01-0.05. The speed regulator performs speed closed-loop regulation according to the speed instruction output by the position controller and the motor speed feedback (e.g., obtained through a Hall sensor or encoder speed calculation). The PID parameters can include Kp=0.8-1.5, Ki=50-120, and Kd=0.005-0.02 to ensure that the thrust component has a smooth response characteristic between high-speed movement and fine positioning, and its speed decays to below 2 mm / s when reaching the target position to improve positioning accuracy. The current driver controls the motor winding current according to the output signal of the speed regulator, and can limit the phase current within a safe range (e.g., 1.5 A-3.0 A) to generate the required electromagnetic torque to drive the motor to rotate, while avoiding coil overheating due to excessive current. The protection circuit monitors the motor working state in real time, and automatically triggers overload protection and outputs a fault alarm signal when the current exceeds the set threshold (e.g., >3.2 A for 200 ms), the motor temperature rise exceeds 80°C, or abnormal conditions such as no response of the motor or loss of encoder pulses are detected. The communication interface can use RS485 or CAN bus to maintain a communication refresh cycle of 10 ms-20 ms with the upper control unit, and is configured to transmit target position instructions, feedback real-time position and speed data, so that the driving motor can respond to the wrinkle position output by the visual recognition module within milliseconds, thereby achieving precise positioning and dynamic following control of the thrust component.
[0098] The machine vision-based flexible film dynamic wrinkle removal device of the present application realizes real-time detection, accurate positioning and dynamic following processing of film wrinkles by integrating a high-speed linear array camera, intelligent image processing algorithms, a flexible flattening roller and a precision thrust actuator. The device can break through the limitations of traditional static wrinkle removal methods, comprehensively monitor the film surface state through the double-sided detection configuration of the visual recognition module, achieve pixel-level wrinkle position recognition accuracy through AI positioning algorithms and coordinate conversion technology. The local arc deformation mechanism of the flexible flattening roller can produce a spatially selective stretching force distribution, realizing non-contact local wrinkle removal processing and avoiding surface damage or marks on the film material. The thrust actuator realizes real-time tracking and processing of dynamic wrinkles through the controllable thrust output of the universal ball adaptive contact and push cylinder, combined with the precise positioning control of the upper and lower slide rail guiding system and the driving motor.
[0099] The closed-loop control logic of the control unit can significantly improve the wrinkle removal accuracy and processing efficiency through the complete automation sequence of image acquisition, wrinkle detection, AI positioning, coordinate conversion, push positioning and deformation wrinkle removal. At the same time, it can adaptively adjust the thrust parameters according to the wrinkle severity, provide personalized wrinkle removal processing solutions for different types and thicknesses of film materials, and minimize the stress damage risk to the film material while ensuring the wrinkle removal effect.
[0100] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.
Claims
1. A flexible thin film dynamic following wrinkle removal device based on machine vision, characterized in that, include: The visual recognition module is used to acquire real-time images of the flexible film in operation and identify the location of wrinkles; A flexible flattening roller, made of flexible material, is used to contact the flexible film; A thrust actuator is disposed on the back side of the flexible flattening roller and includes a thrust component that can move along the axial direction of the flexible flattening roller. A moving device for driving the thrust component to move axially to the folded position; as well as The control unit is used to control the thrust actuator according to the wrinkle position identified by the vision recognition module, so that the thrust component is aligned with the wrinkle position and pushes the flexible flattening roller to produce local deformation, thereby forcibly flattening the film wrinkles.
2. The flexible thin film dynamic following wrinkle removal device based on machine vision according to claim 1, characterized in that, The visual recognition module includes a high-speed linear array camera for real-time acquisition of linear array images of the flexible thin film in operation.
3. The machine vision-based flexible film dynamic following wrinkle removal device according to claim 2, characterized in that, The control unit has a built-in image processing algorithm for real-time segmentation and analysis of the linear array image, and for identifying the position and width of the wrinkles.
4. The machine vision-based flexible film dynamic following wrinkle removal device according to claim 3, characterized in that, The image processing algorithm includes: The acquired linear array images are preprocessed, including noise filtering, contrast enhancement, and image correction. The grayscale changes and texture features of the thin film surface are identified using an edge detection algorithm; The geometric center of the wrinkled region is determined by centroid calculation, and the pixel position of the wrinkle in the image coordinate system is converted into the physical position in the actual width direction of the film. The boundary contour of the folded region is analyzed using contour analysis methods to calculate the effective width range of the folds. Based on the multidimensional feature parameters of the geometric shape, area size, aspect ratio and gray level distribution of the folded region, the system uses a preset feature template to determine whether the detected region is a real fold. It outputs precise information on the position and width of the folds, providing accurate parameter data for the precise positioning of the thrust actuator and the local deformation control of the flexible flattening roller.
5. The flexible thin film dynamic following wrinkle removal device based on machine vision according to claim 1, characterized in that, The flexible flattening roller is made of soft rubber material and is flexible.
6. The machine vision-based flexible thin film dynamic following wrinkle removal device according to claim 5, characterized in that, The control system for the drive motor includes: The position controller receives the target position command from the control unit, obtains the real-time position information of the drive motor through encoder feedback, calculates the position deviation, and outputs the speed command. The speed regulator is used to adjust the running speed of the motor according to the speed command of the position controller and the motor speed feedback signal, through a PID algorithm, to ensure that the thrust component moves to the target position at an appropriate speed. A current driver is used to receive the output signal of the speed regulator, control the magnitude and direction of the motor winding current, and generate corresponding torque to drive the motor to rotate. The protection circuit is used to monitor the operating status of the motor and provide overload protection and fault alarm in abnormal conditions; The communication interface is used to realize data exchange and command transmission between the control system and the upper control unit, ensuring that the drive motor can accurately respond to the wrinkle detection results and realize precise positioning control of the thrust component when receiving the wrinkle position command.
7. The flexible thin film dynamic following wrinkle removal device based on machine vision according to claim 1, characterized in that, The mobile device includes: Upper and lower slide rails are used to guide the axial movement of the thrust component; and A drive motor is used to drive the thrust component to move along the upper slide rail and the lower slide rail.
8. The flexible thin film dynamic following wrinkle removal device based on machine vision according to claim 1, characterized in that, The thrust actuator also includes: A swivel ball is used to accommodate the arcuate deformation of the flexible flattening roller; and The cylinder is pushed so that it contacts the flexible flattening roller through the omnidirectional ball and applies a thrust.
9. The flexible thin film dynamic following wrinkle removal device based on machine vision according to claim 1, characterized in that, It also includes a front wrinkle visual recognition module and a back wrinkle visual recognition module, which are used to detect wrinkles on the front and back of the flexible film, respectively.
10. The flexible thin film dynamic following wrinkle removal device based on machine vision according to claim 1, characterized in that, The control unit converts the pixel coordinates of the folds into mechanical coordinates to achieve real-time positioning control of the thrust component.
Citation Information
Patent Citations
Wrinkle recognition method and apparatus for capacitor thin film
CN106529510A
Local tensioning and flattening device for flexible base material
CN115924598A
Extending apparatus for flexible film
JP2012099606A
Wrinkle removal system
JP2025012936A
Cited By
Self-adaptive folding method and system for image bag taking based on machine learning
CN121757647A
Machine learning-based adaptive folding method and system for image bag taking
CN121757647B