Visual inspection method and inspection equipment for rubber tube finished product

Abstract

The invention relates to the technical field of rubber tube detection, and discloses a rubber tube finished product visual detection method and detection device.The method comprises the following steps that rubber tubes are fixed in a first limiting plate, a second limiting plate and a third limiting plate on a limiting assembly of a detection table; controlling the manipulator to carry the visual lens to move to a preset detection position; based on the real-time position coordinates of the manipulator and the to-be-detected feature type, triggering a corresponding multi-band light source: when the to-be-detected feature type is a metal clamp or a pipe clamp, starting a polarized visible light source to inhibit overexposure. The optimal light source can be automatically selected according to the to-be-detected feature type, and the system triggers the polarized visible light source according to the metal clamp or pipe clamp features, so that light reflection is inhibited; the near-infrared transmission light source is triggered according to the characteristics of a clamp spring or a rubber check ring, and the activation time period of the polarized visible light source and the infrared transmission light source is synchronous with the exposure signal of the camera, and only works in an exposure window.

Description

Technical Field

[0001] This invention relates to the field of rubber hose inspection technology, specifically to a visual inspection method and equipment for finished rubber hoses. Background Technology

[0002] In the automotive manufacturing industry, rubber hoses are the core components of fluid transmission systems. The quality of rubber hose assembly determines the safety and sealing reliability of the entire machine. To ensure the quality of rubber hoses, it is necessary to conduct comprehensive quality control on finished rubber hoses equipped with metal clamps, pipe clamps, spring clips, and rubber retaining rings. Advanced visual inspection technology is used to replace traditional manual visual inspection to accurately identify missing components, positional deviations, surface cracks, and internal structural defects.

[0003] In existing rubber hose finished product inspection processes, operators manually place the rubber hose into a fixed fixture on the inspection table for positioning. After the equipment is started, an industrial camera mounted on a fixed bracket, along with a constantly lit single-band light source, photographs the surface of the rubber hose. The acquired image data is transmitted to a central processing unit, which uses preset grayscale thresholds to determine the presence of defects in each component or on the surface. After the inspection is completed, the operator manually unlocks the fixture and removes the product. Subsequently, the product is manually sorted into different collection boxes based on the inspection results displayed on the screen, completing the entire inspection cycle.

[0004] Currently, the finished rubber tubes to be tested contain both metal clamps with high specular reflection characteristics and rubber material with low reflectivity that needs to be observed through transmission. A single mode of visible light source cannot simultaneously meet the imaging needs of both. During testing, the metal surface may form light spots due to strong reflection, which may cover up fine cracks, or the rubber retaining ring may not show internal bubbles and delamination characteristics due to insufficient light penetration. This can easily lead to missed detections or misjudgments.

[0005] Therefore, the purpose of this invention is to provide a visual inspection method and equipment for finished rubber hoses, so as to overcome the shortcomings of the prior art. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a visual inspection method and equipment for finished rubber tubes, which solves the problem that currently, finished rubber tubes to be inspected simultaneously contain metal clamps with high specular reflection characteristics and rubber material with low reflectivity that requires transmission observation, and a single mode of visible light source cannot simultaneously meet the imaging requirements of both.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] The first aspect of this invention provides a visual inspection method for finished rubber hoses, comprising the following steps:

[0009] Step S1: Fix the rubber tube inside the first limiting plate, the second limiting plate and the third limiting plate on the limiting assembly of the testing table;

[0010] Step S2: Control the robotic arm to move the vision lens to the preset detection position;

[0011] Step S3: Based on the real-time position coordinates of the robotic arm and the type of feature to be measured, trigger the corresponding multi-band light source:

[0012] When the type of feature to be measured is a metal clamp or pipe clamp, a polarized visible light source is enabled to suppress overexposure.

[0013] When the type of feature to be tested is a snap ring or a rubber retainer, the near-infrared transmission light source is activated.

[0014] Step S4: Acquire multi-angle images of the rubber tube at different locations using a visual lens, and generate a comprehensive detection image through weighted fusion;

[0015] Step S5: Control the first cylinder, the second cylinder, and the third cylinder to retract the first limit plate, the second limit plate, and the third limit plate. Based on the detection result of the weighted fusion generating comprehensive detection image, control the forward and reverse conveyor belt to sort the detected rubber blocks into qualified or unqualified product frames.

[0016] Preferably, the control parameters of the multi-band light source include the visible light band and the near-infrared band, and the visible light band supports polarization state switching.

[0017] Preferably, the polarization state switching specifically includes:

[0018] Polarized light sources are automatically activated for metal clamps or pipe clamps, and the light source is triggered only when the robot is in the preset detection position to reduce ineffective light energy consumption. Non-polarized visible light sources are activated for non-metallic areas.

[0019] Preferably, the triggering condition for the near-infrared light source in step S3 is:

[0020] When the type of feature to be tested is a snap ring or a rubber retainer, a near-infrared transmission light source with a wavelength range of 850-950nm is activated, and the near-infrared light source is triggered only when the robot is located in the feature space interval, where the feature space interval is the coordinate of a preset detection position.

[0021] Preferably, the weighted fusion to generate the comprehensive detection image includes:

[0022] Based on the images from each acquisition location and the corresponding light source band parameters, weighting coefficients are assigned and then superimposed to generate a comprehensive detection image.

[0023] Preferably, the specific determination process for the detection result of the weighted fusion-generated comprehensive detection image in step S5 is as follows:

[0024] The feature parameters of the target feature type in the weighted fusion generated comprehensive detection image are compared with the pre-stored standard template to calculate the difference value.

[0025] If the difference value is less than the preset tolerance threshold, the test result is deemed qualified.

[0026] If the difference value exceeds the preset tolerance threshold, the test result is determined to be unqualified.

[0027] Preferably, the multi-band light source is integrated into the end effector of the robotic arm, and the multi-band light source is coaxially mounted with the vision lens. The multi-band light source includes a visible light unit with a wavelength range of 400-700nm and a linear polarizer driven by a stepper motor, and a near-infrared transmission unit with a wavelength range of 850-950nm.

[0028] Preferably, the vision lens is equipped with a position sensor, which is used to acquire the coordinates of the vision lens in real time.

[0029] A second aspect of the present invention provides a visual inspection device for finished rubber hoses, comprising a main body of the device, an inspection platform provided on the main body of the device, and a limiting component provided on the inspection platform. The limiting component includes a first cylinder, a second cylinder and a third cylinder, wherein the first cylinder is used to drive a first limiting plate, the second cylinder is used to drive a second limiting plate, and the third cylinder is used to drive a third limiting plate.

[0030] Preferably, the main body of the device is equipped with a controller, the top inner wall of the main body of the device is fixedly connected to a robotic arm, the robotic arm is equipped with a vision lens, the side wall of the main body of the device is equipped with a grating, the front surface of the main body of the device is equipped with a start switch, a conveyor belt is provided below the inspection table, and defective product frames and qualified product frames are provided on both sides of the conveyor belt. The controller integrates a communication interface module, an input / output control module, a light source driving module, a main control logic module, a result judgment module, a motion control module, and an image processing system. The image processing system includes a preprocessing module, a weight calculation module, an image registration and fusion module, and a defect analysis module.

[0031] This invention provides a visual inspection method and equipment for finished rubber hoses. It has the following beneficial effects:

[0032] 1. This invention sets the linkage between the position of the robotic arm end effector and the light source mode. When the robotic arm reaches the preset spatial coordinates, the optimal light source is automatically selected according to the type of feature to be measured. For metal clamps or pipe clamps, the system triggers a polarized visible light source to suppress reflection and eliminate light spots that cover up fine cracks. For snap rings or rubber retaining rings, the system triggers a near-infrared transmission light source to penetrate the rubber material and clearly show internal bubbles and delamination features, thereby improving the detection quality.

[0033] 2. The activation period of the polarized visible light source and infrared transmission light source of the present invention is synchronized with the camera exposure signal, and they only work within the exposure window, thereby reducing the ineffective energy consumption of the polarized visible light source and infrared transmission light source.

[0034] 3. This invention integrates multi-angle images acquired by the vision lens at different locations using a weighted fusion method based on the acquisition location and light source wavelength parameters. Weighting coefficients are assigned according to the spatial coordinates of each acquisition location and the corresponding light source wavelength parameters. A comprehensive detection image is generated through superposition calculation, simultaneously considering both high reflectivity and transmission imaging, thus improving the imaging clarity of the metal clamp edge and the internal structure of the rubber retaining ring.

[0035] 4. This invention controls the running direction of the conveyor belt based on the detection results. After the detection is completed, under the action of the limiting component, the output ends of the first cylinder, the second cylinder, and the third cylinder are synchronously driven to extend and move back, and the detection part is no longer limited. The system drives the conveyor belt to perform forward or reverse conveying according to the qualified or unqualified judgment result, so as to ensure the sorting and positioning accuracy. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the overall structure of the present invention;

[0037] Figure 2 This is a schematic diagram of the robotic arm structure of the present invention;

[0038] Figure 3 This is a schematic diagram of the limiting component structure of the present invention;

[0039] Figure 4 This is a schematic diagram of the image processing system workflow of the present invention;

[0040] Figure 5 This is a schematic diagram of the internal logic control of the controller of the present invention;

[0041] Figure 6 This is a schematic diagram of the controller signal output and component control of the present invention.

[0042] The components include: 1. Equipment body; 2. Controller; 3. Robotic arm; 4. Grating; 5. Start switch; 6. Non-conforming product frame; 7. Conforming product frame; 8. Conveyor belt; 9. Vision lens; 10. Inspection table; 11. Limiting assembly; 1101. First cylinder; 1102. First limiting plate; 1103. Second cylinder; 1104. Second limiting plate; 1105. Third cylinder; 1106. Third limiting plate; 12. Preprocessing module; 13. Weight calculation module; 14. Image registration and fusion module; 15. Defect analysis module; 16. Result judgment module; 17. Motion control module; 18. Communication interface module; 19. Input / output control module; 20. Light source drive module; 21. Main control logic module. Detailed Implementation

[0043] The technical solutions in 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.

[0044] See attached document Figures 1 to 3 This invention provides a specific embodiment of a visual inspection device for finished rubber hoses. The visual inspection device for finished rubber hoses includes a main body 1, a controller 2, a robotic arm 3, a grating 4, a start switch 5, a defective product frame 6, a qualified product frame 7, a conveyor belt 8, a vision lens 9, an inspection table 10, and a limiting assembly 11. The limiting assembly 11 includes a first cylinder 1101, a first limiting plate 1102, a second cylinder 1103, a second limiting plate 1104, a third cylinder 1105, and a third limiting plate 1106.

[0045] The main body 1 of the finished rubber hose visual inspection equipment is the basic frame. The controller 2 is installed on the surface of the main body 1 to coordinate the operation of the components. The robotic arm 3 is fixed to the inner top wall of the main body 1, and the end of the robotic arm 3 integrates a vision lens 9 and a multi-band light source unit. The grating 4 is set on the side wall of the main body 1 for safety monitoring of the working area. The start switch 5 is located on the front surface of the main body 1 to control the start and stop of the system.

[0046] The testing platform 10 is located in the operating area of ​​the main body 1 of the equipment and is used to support the rubber tube to be tested. A limiting assembly 11 is installed on the surface of the testing platform 10. A first cylinder 1101 drives a first limiting plate 1102, a second cylinder 1103 drives a second limiting plate 1104, and a third cylinder 1105 drives a third limiting plate 1106, achieving three-dimensional positioning of the rubber tube. A conveyor belt 8 is located below the testing platform 10 to receive the tested rubber tubes. Defective product frames 6 and qualified product frames 7 are located on opposite sides of the conveyor belt 8 in the conveying direction. In use, the rubber tube to be tested is placed in the limiting assembly 11 of the testing platform 10. The first cylinder 1101, second cylinder 1103, and third cylinder 1105 drive the first limiting plate 1102, second limiting plate 1104, and third limiting plate 1106 to work together to fix the rubber tube. The controller 2 then controls the robotic arm 3, carrying the vision lens 9, to move to the preset testing position. The multi-band light source unit at the end of the robotic arm 3 triggers the corresponding light source according to the current position coordinates and the type of feature to be measured. Polarized visible light source is activated for metal clamps or pipe clamps; near-infrared transmission light source is activated for snap rings or rubber retaining rings. The vision lens 9 acquires multi-angle images at each detection position.

[0047] The acquired images are transmitted to controller 2, where they undergo weighted fusion processing based on position information and light source parameters to generate a comprehensive detection image. After detection, the first cylinder 1101, the second cylinder 1103, and the third cylinder 1105 retract the first limit plate 1102, the second limit plate 1104, and the third limit plate 1106, respectively. The rubber tube falls onto the conveyor belt 8, and controller 2 controls the direction of the conveyor belt 8 according to the detection results: qualified products are conveyed forward to the qualified product frame 7; unqualified products are conveyed in reverse to the unqualified product frame 6. The grating 4 monitors the working area in real time, triggering a shutdown protection mechanism in case of abnormal conditions.

[0048] See attached document Figure 3 This is a limiting component 11 for a visual inspection device for finished rubber hoses. The limiting component 11 includes a first cylinder 1101, a first limiting plate 1102, a second cylinder 1103, a second limiting plate 1104, a third cylinder 1105, and a third limiting plate 1106. The first cylinder 1101, the first limiting plate 1102, the second cylinder 1103, the second limiting plate 1104, the third cylinder 1105, and the third limiting plate 1106 are all mounted on the surface of the inspection table 10. The first cylinder 1101 is connected to the first limiting plate 1102 via a rigid connecting rod, the second cylinder 1103 is connected to the second limiting plate 1104 via a rigid connecting rod, and the third cylinder 1105 is connected to the third limiting plate 1106 via a rigid connecting rod.

[0049] After the test is completed, the controller releases the fixing of the test piece, and the controller 2 sends a command simultaneously. The first cylinder 1101 drives the first limit plate 1102 to return to the initial position. The second cylinder 1103 drives the second limit plate 1104 to return to the initial position. The third cylinder 1105 drives the third limit plate 1106 to return to the initial position, and the rubber tube of the test piece falls freely onto the conveyor belt 8.

[0050] See attached document Figures 1 to 3 The visual inspection unit provided by this invention consists of a robotic arm 3, a vision lens 9, and a multi-band light source unit. The robotic arm 3 is an industrial robot, and its base is fixed to the top inner wall of the main body 1 by bolts. The end flange of the robotic arm 3 is provided with an installation interface, and the vision lens 9 and the multi-band light source unit are coaxially mounted through the interface, with an axial distance of 15mm.

[0051] The Vision Lens 9 utilizes a global shutter CMOS industrial camera with a 20-megapixel sensor. The lens features a motorized zoom mechanism with a focal length adjustment range of 35mm to 100mm. The multi-band light source unit comprises two independent light source modules: a visible light source module with a wavelength range of 400-700nm, and a near-infrared transmission light source module with a wavelength range of 850-950nm. The visible light source module incorporates a linear polarizer driven by a stepper motor, with a polarization angle... Defined by physical formulas: This formula characterizes the polarizer angle. The relationship between (unit: degrees) and extinction ratio, where: Indicates the polarization extinction ratio. This indicates the rotation angle of the polarizer.

[0052] Controller 2 stores a set of spatial coordinates for preset detection positions:

[0053] ;

[0054] in: Represents a preset set of position coordinates; Indicates the first Three-dimensional coordinates of each detection location (in millimeters); This indicates the total number of preset positions.

[0055] The coordinate values ​​were obtained through calibration using a coordinate measuring machine. The robot arm 3 uses the DH parameter method to establish its kinematic model, and the end-effector pose transformation matrix is ​​expressed as:

[0056] ;

[0057] in: Represents the homogeneous transformation matrix of the end effector; Indicates the first Transformation matrix of each joint; This represents the rotation angle of each joint. The 16-bit absolute encoder built into the vision lens 9 outputs the position deviation in real time.

[0058] ;

[0059] in: Represents the position deviation vector (in millimeters); Indicates the actual position coordinates; The coordinates represent the target location.

[0060] When the L2 norm of the positional deviation At that time, the positioning is determined to be complete.

[0061] See attached document Figure 4 Controller 2 queries the feature type mapping function based on the location coordinates. Establish position coordinates With feature set Metal clamps, pipe clamps, retaining rings, rubber retaining rings The correspondence between them, where, This represents a mapping function from spatial coordinates to feature types. Light source triggering control is executed based on this function.

[0062] ;

[0063] in, This represents the light source control strategy function. Polarization mode setting. Optimize the suppression of reflections on metal surfaces. Near-infrared mode output power is set to 50W.

[0064] The light source and camera exposure are strictly synchronized in time:

[0065] ;

[0066] ;

[0067] in: Indicates the light source on time (in milliseconds); Indicates the time the light source is off (in milliseconds); Indicates the start time of exposure; Indicates the end time of the exposure.

[0068] The visible light source module has a fixed illumination duration of 10ms, while the near-infrared transmission light source module operates in a 5ms single-pulse mode. Three images are acquired at each detection location, and the image sequence is defined as follows:

[0069] ;

[0070] in Represents a set of images; This indicates the total number of images.

[0071] Each image's attached metadata record contains the following fields:

[0072] Spatial coordinate data: 3D floating-point vector (Unit: millimeters);

[0073] Feature type identifier: Integer enumeration value (1: metal clamp, 2: pipe clamp, 3: snap ring, 4: rubber retaining ring);

[0074] Polarization angle value: Floating-point data (unit: degrees);

[0075] Light source mode identifier: Boolean data (0: visible light mode, 1: near-infrared mode);

[0076] Exposure time parameter: Integer data (in microseconds);

[0077] The data structure is used for parameter correlation in subsequent image processing stages to ensure that the light source parameters at each detection location correspond accurately to the image data.

[0078] To achieve high-precision feature recognition, the image processing system is specifically divided into a preprocessing module 12, a weight calculation module 13, an image registration and fusion module 14, and a defect analysis module 15. The modules transmit image data and associated metadata through an internal data bus.

[0079] The preprocessing module 12 is configured to receive raw image sequence data from the vision lens 9. The image sequence consists of multiple images acquired during the movement of the robotic arm along a preset path. The metadata of each image precisely records the spatial coordinates of the robotic arm's end effector at the moment of acquisition, the light source wavelength parameters (such as visible light or near-infrared), and the corresponding feature type identifier. The preprocessing module 12 first performs format standardization processing on the image data. For data acquired using a color image sensor, the preprocessing module 12 performs grayscale conversion. The conversion process follows the brightness calculation standard recommended by the International Telecommunication Union (ITU), which assigns weight coefficients based on the differences in human eye sensitivity to the wavelengths of red, green, and blue colors. In specific implementation, the pixel values ​​of the red, green, and blue channels are multiplied by specific coefficients (usually 0.299, 0.587, and 0.114, respectively). The algebraic sum of these three products is the converted single-channel grayscale pixel value, thereby reducing the three-dimensional color space to a one-dimensional brightness space.

[0080] After grayscale conversion, the preprocessing module 12 further performs noise reduction on the image. To suppress noise while preserving edge details of the rubber tube surface, the module employs a Gaussian filtering algorithm. The Gaussian filtering algorithm performs a sliding window operation on the image matrix using a convolution kernel of a preset size (e.g., 5×5 pixels). The numerical distribution of the convolution kernel follows a two-dimensional Gaussian normal distribution, meaning the weight is highest at the center point, and decreases with increasing distance from the center point in a bell-shaped curve. Through this weighted averaging operation, high-frequency random noise in the image is smoothly filtered out, providing a high-quality data foundation for subsequent feature extraction.

[0081] The weight calculation module 13 is configured to calculate the normalized weight coefficient for each image based on the acquisition conditions. The weight calculation module 13 receives the preprocessed image set along with relevant spatial coordinates and light source parameters, and employs a two-factor evaluation strategy combining position weight and band weight. For position weight calculation, the weight calculation module 13 uses a Gaussian attenuation model based on Euclidean distance. Using the optimal observation position coordinates calibrated by the model as the center, the spatial distance between the current image acquisition position and the optimal position is calculated. According to the mathematical properties of the Gaussian function, the smaller the distance, the larger the calculated position weight factor; conversely, as the distance increases, the weight factor decays exponentially. This mechanism ensures that images captured at the optimal viewing angle dominate the fusion result.

[0082] In terms of band weight calculation, the weight calculation module 13 assigns values ​​according to a preset feature-light source matching logic table. The logic table defines the imaging advantages of specific features under specific light source bands. For example, when the metadata indicates that the current detection area is a metal clamp and the light source is in polarization mode, the system assigns a high band weight factor to the image because polarized light can significantly suppress specular reflection on the metal surface; when the detection area is a rubber spring clip and the light source is in near-infrared transmission mode, the system assigns a high band weight factor by utilizing the penetration characteristics of near-infrared light through rubber materials. For cases that do not meet the above specific advantage combinations, a basic weight factor is assigned. Subsequently, the weight calculation module 13 linearly weights the position weight factor and the band weight factor according to a preset ratio, and normalizes the comprehensive weight of all images participating in the fusion process to ensure that the sum of the final weight coefficients of all images is strictly equal to 1, ensuring that the brightness level of the fused image is consistent with the original image.

[0083] The image registration and fusion module 14 is configured to solve the spatial alignment and data overlay problems of multi-view images. Since the robotic arm 3 captures images from different positions, resulting in perspective differences, the image registration and fusion module 14 first uses the Scale Invariant Feature Transform (SIFT) algorithm to extract key feature points from each image. These feature points are rotation-, scale-, and translation-invariant. The module uses a feature descriptor matching algorithm to find corresponding point pairs between different images and calculates the homography transformation matrix based on these point pairs. Using the matrix, the module transforms and projects all images onto the same reference coordinate system, eliminating perspective bias. After completing spatial registration, the module performs pixel-level weighted fusion. For each pixel coordinate in the reference coordinate system, the module reads the pixel value of each registered image at that point and combines it with the corresponding normalized weight coefficients output by the weight calculation module 13 to perform a weighted summation operation. To prevent data overflow and precision loss during the calculation process, the fusion result is stored in a 32-bit floating-point data format.

[0084] The defect analysis module 15 is configured to extract and determine features from the fused high dynamic range image. The module uses the Sobel edge detection operator to perform convolution operations on the image to extract the defect contours of the rubber tube surface. The Sobel operator includes two templates, one horizontal and one vertical, which calculate the horizontal and vertical gradients of the image brightness, respectively. The gradient magnitude at any point in the image is determined by the vector sum of these two gradients. The defect analysis module 15 compares the calculated gradient magnitudes with a preset threshold; regions exceeding the threshold are identified as edges or defect areas. By further combining morphological closing operations to fill edge breakpoints, the module can construct a complete defect contour and determine whether the assembly quality of the rubber tube assembly is acceptable based on the geometric dimensions of the contour.

[0085] See attached document Figure 5 The detection result determination and sorting control process provided by this invention is mainly completed collaboratively by the result determination module 16 and motion control module 17 integrated within the controller 2. The process immediately follows the image processing and fusion step, aiming to achieve automatic grading and physical sorting of products based on the fused comprehensive image data.

[0086] The result determination module 16 is configured to receive comprehensive inspection image data from the preceding module. Internally, the result determination module 16 stores feature parameter templates for standard qualified products and defect determination thresholds. The result determination module 16 performs feature comparison calculations on the comprehensive inspection image, calculating the difference between the feature parameters of each key region in the image and the standard template. When the calculated difference value is less than a preset tolerance threshold, the result determination module 16 generates a Boolean status signal indicating qualification; when the difference value exceeds the preset tolerance threshold, the result determination module 16 generates a Boolean status signal indicating non-qualification. The status signal is directly transmitted to the motion control module 17 as a sorting instruction.

[0087] The motion control module 17 is configured to coordinate the actions of external actuators based on received status signals. Upon receiving a status signal, the motion control module 17 first executes the release action logic. The motion control module 17 sends a synchronous retraction command to the first cylinder 1101, the second cylinder 1103, and the third cylinder 1105 via the I / O interface. In response to the command, the piston rods of the first cylinder 1101, the second cylinder 1103, and the third cylinder 1105 simultaneously retract into the cylinder.

[0088] As the piston rods of the first cylinder 1101, the second cylinder 1103, and the third cylinder 1105 retract, the first limiting plate 1102, the second limiting plate 1104, and the third limiting plate 1106 move radially outward. This action releases the spatial constraint of the limiting assembly 11 on the rubber tube located at the center of the detection area. Under the action of gravity, the rubber tube leaves its original suspended position and falls vertically onto the surface of the conveyor belt 8 located directly below.

[0089] The motion control module 17 simultaneously controls the running direction of the drive motor of the conveyor belt 8 based on the content of the status signal. When the received status signal is acceptable, the motion control module 17 outputs a positive voltage signal to the drive motor. The conveyor belt 8 is set to a positive speed. Operation. The forward direction of operation points spatially to the qualified product frame 7 located on the right side of the equipment. The rubber tube is conveyed to the right by the friction of the belt and eventually slides into the qualified product frame 7 for collection.

[0090] When the received status signal is unqualified, the motion control module 17 outputs a reverse voltage signal to the drive motor. The conveyor belt 8 switches to reverse speed. Operation. The reverse direction of operation points spatially towards the defective product frame 6 located on the left side of the equipment. Driven by belt friction, the rubber tube is conveyed to the left and eventually slides into the defective product frame 6. Through this logic, the system achieves automated physical sorting based on visual inspection results, ensuring the physical separation of qualified and defective products.

[0091] See attached document Figure 6The vision inspection system provided by this invention uses a controller 2 as the core control unit, which establishes electrical connections and data interactions with external actuators and sensing devices through an integrated hardware interface module. The controller 2 adopts an industrial control computer based on the x86 architecture, and runs a real-time operating system to ensure the determinism of task scheduling. To achieve centralized management and hierarchical control of the system, the controller 2 is internally configured with a communication interface module 18, an input / output control module 19, a light source drive module 20, and a main control logic module 21.

[0092] The communication interface module 18 is configured to enable communication between the controller 2 and high-bandwidth data devices and intelligent terminals. The communication interface module 18 integrates an EtherCAT bus master interface, which connects to the servo drive of the robot arm 3 via an industrial Ethernet cable. The controller 2 sends joint angle commands or Cartesian space coordinate commands through the interface and reads the position feedback data of each joint encoder of the robot arm 3 in real time at a 1ms cycle. In addition, the communication interface module 18 also integrates a GigEVision gigabit Ethernet port, which connects to the vision lens 9 via a Cat6 network cable. The interface is responsible for transmitting camera trigger commands, configuring exposure parameters, and receiving high-resolution raw image data streams.

[0093] The input / output control module 19 is configured to process digital and analog signals in the system. The module connects to the safety light curtain 4 and the start switch 5 via opto-isolated digital input ports to monitor input levels in real time and determine operator commands and on-site safety conditions. The digital output ports of the input / output control module 19 are connected to the solenoid valve group that drives the first cylinder 1101, the second cylinder 1103, and the third cylinder 1105 via a relay array, outputting a 24V voltage signal to control the extension and retraction of the cylinder piston rods. The input / output control module 19 also includes a motor drive interface connected to the driver of the conveyor belt 8, controlling the start / stop and running direction of the conveyor belt 8 by outputting positive / negative voltage or pulse direction signals.

[0094] The light source driver module 20 is configured for fine-grained control of the multi-band light source unit. The light source driver module 20 features multi-channel digital I / O output and PWM pulse width modulation output. The controller 2 sends flicker trigger signals to the visible light source module and the near-infrared transmission light source module through the light source driver module 20, with signal timing accuracy controlled at the microsecond level. Simultaneously, the duty cycle of the PWM signal output by the light source driver module 20 determines the instantaneous luminous intensity of the light source, thereby meeting the differentiated light intensity requirements of different material characteristics.

[0095] The main control logic module 21 is the core program unit running in the memory of the controller 2, responsible for coordinating the hardware modules to execute the detection steps S1 to S5 as described in claim 1. After the system is powered on, the main control logic module 21 enters the standby monitoring state. When the rising edge signal generated by the start switch 5 is detected by the input / output control module 19, the main control logic module 21 first queries the input status of the safety light curtain 4. If the safety light curtain 4 is in the blocked state, the system remains stopped and reports an error; if the safety light curtain 4 is in the conducting state, indicating that the detection area is safe, the main control logic module 21 then drives the solenoid valve group to act through the output port, controlling the first cylinder 1101, the second cylinder 1103 and the third cylinder 1105 to extend the piston rod synchronously, driving the limit assembly 11 to fix the rubber tube to be tested at the detection center position.

[0096] After the rubber tube is fixed, the main control logic module 21 starts the detection cycle. Based on the preset set of detection points, the main control logic module 21 sends the target position coordinate command to the robot arm 3 via the communication interface module 18. The robot arm 3 executes its motion plan and reaches the designated spatial position. The main control logic module 21 reads the current position coordinates and, based on the feature mapping relationship... The required light source mode is determined. Subsequently, the main control logic module 21 sends a configuration command to the light source driver module 20 to set the polarization or near-infrared mode and the corresponding luminous power, and sends a hard trigger signal to the vision lens 9.

[0097] After completing image acquisition and receiving image data, the main control logic module 21 calls the aforementioned image processing system (including the preprocessing module 12, weight calculation module 13, image registration and fusion module 14, and defect analysis module 15) to perform calculations on the image sequence and finally obtain the detection result identifier. Based on the test results, the main control logic module 21 executes the sorting logic. First, it de-energizes the solenoid valve group, causing all cylinders to retract and releasing the rubber hose. Then, if the test results... To ensure compliance, the main control logic module 21 controls the conveyor belt 8 to run in the forward direction, transporting the rubber tube to the qualified product frame 7; if the test result... If the item is found to be defective, the conveyor belt 8 is reversed to transport the rubber tube to the defective item frame 6. Through the combination of the above hardware architecture and software logic, fully automated detection and sorting are achieved.

Claims

1. A method of visual inspection of a finished rubber tube, characterized in that, Includes the following steps: Step S1: Fix the rubber tube inside the first limiting plate (1102), the second limiting plate (1104) and the third limiting plate (1106) on the limiting assembly (11) of the testing table (10); Step S2: Control the robotic arm (3) to move the vision lens (9) to the preset detection position; Step S3: Based on the real-time position coordinates of the robotic arm (3) and the type of feature to be measured, trigger the corresponding multi-band light source: When the type of feature to be measured is a metal clamp or pipe clamp, a polarized visible light source is enabled to suppress overexposure. When the type of feature to be tested is a snap ring or a rubber retainer, the near-infrared transmission light source is activated. Step S4: Collect multi-angle images of the rubber tube at different positions through the visual lens (9), and generate a comprehensive detection image through weighted fusion; Step S5: Control the first cylinder (1101), the second cylinder (1103), and the third cylinder (1105) to retract the first limit plate (1102), the second limit plate (1104), and the third limit plate (1106). Based on the detection results of the weighted fusion generated comprehensive detection image, control the conveyor belt (8) to transport the detected rubber blocks in the forward and reverse directions to sort them into qualified product frames (7) or unqualified product frames (6).

2. The method of claim 1, wherein The control parameters of the multi-band light source include the visible light band and the near-infrared band, and the visible light band supports polarization state switching.

3. The method of claim 2, wherein the method further comprises: The polarization state switching specifically includes: The polarized light source is automatically activated for metal clamps or pipe clamps, and the light source is triggered only when the robot (3) is in the preset detection position to reduce the energy consumption of ineffective lighting. The non-polarized visible light source is activated for non-metallic areas.

4. The method of claim 1, wherein The triggering condition for the near-infrared light source in step S3 is as follows: When the type of feature to be tested is a snap ring or a rubber retainer, a near-infrared transmission light source with a wavelength range of 850-950nm is activated, and the near-infrared light source is triggered only when the robot (3) is in the feature space interval, where the feature space interval is the coordinate of the preset detection position.

5. The method of claim 1, wherein the method further comprises: The weighted fusion to generate the comprehensive detection image includes: Based on the images from each acquisition location and the corresponding light source band parameters, weighting coefficients are assigned and then superimposed to generate a comprehensive detection image.

6. The visual inspection method for finished rubber hoses according to claim 1, characterized in that, The specific determination process of the detection result in step S5 based on the weighted fusion to generate the comprehensive detection image is as follows: The feature parameters of the target feature type in the weighted fusion generated comprehensive detection image are compared with the pre-stored standard template to calculate the difference value. If the difference value is less than the preset tolerance threshold, the test result is deemed qualified. If the difference value exceeds the preset tolerance threshold, the test result is determined to be unqualified.

7. The method of claim 1, wherein the method further comprises: The multi-band light source is integrated at the end of the robotic arm (3). The multi-band light source is coaxially mounted with the vision lens (9). The multi-band light source includes a visible light unit with a wavelength range of 400-700nm and a linear polarizer driven by a stepper motor, and a near-infrared transmission unit with a wavelength range of 850-950nm.

8. The method of claim 1, wherein, The vision lens (9) is equipped with a position sensor, which is used to obtain the coordinates of the vision lens (9) in real time.

9. A rubber tube finished product visual inspection apparatus characterized by comprising: A visual inspection method for finished rubber tubes according to any one of claims 1-8 includes a main body (1), an inspection table (10) is provided on the main body (1), and a limiting component (11) is provided on the inspection table (10). The limiting component (11) includes a first cylinder (1101), a second cylinder (1103) and a third cylinder (1105). The first cylinder (1101) is used to drive a first limiting plate (1102), the second cylinder (1103) is used to drive a second limiting plate (1104), and the third cylinder (1105) is used to drive a third limiting plate (1106).

10. The rubber tube finished product visual inspection apparatus according to claim 9, wherein The main body (1) of the equipment is equipped with a controller (2). The top inner wall of the main body (1) is fixedly connected to the robot (3). The robot (3) is equipped with a vision lens (9). The side wall of the main body (1) is equipped with a grating (4). The front surface of the main body (1) is equipped with a start switch (5). The bottom of the inspection table (10) is equipped with a conveyor belt (8). The two sides of the conveyor belt (8) are equipped with a defective product frame (6) and a qualified product frame (7). The controller (2) integrates a communication interface module (18), an input / output control module (19), a light source drive module (20), a main control logic module (21), a result judgment module (16), a motion control module (17), and an image processing system. The image processing system includes a preprocessing module (12), a weight calculation module (13), an image registration and fusion module (14), and a defect analysis module (15).

Images