Light guide film cutting and positioning system based on machine vision

Abstract

The invention relates to the technical field of light guide films, in particular to a light guide film cutting positioning system based on machine vision, which comprises a light guide film image processing module, an edge feature extraction module, a cutter adjustment module and a cutting effect evaluation module. Through real-time image data acquisition and edge detection and feature extraction of the convolutional neural network, the accuracy and efficiency of light guide film cutting are remarkably improved, the system is allowed to accurately identify the edge position of the light guide film, errors are reduced, the cutting quality is optimized, a real-time feedback mechanism allows the system to carry out immediate adjustment according to the cutting effect, and the accuracy and efficiency of light guide film cutting are improved. The adaptability and continuity of the cutting operation are improved, the convolutional neural network is optimized in combination with the cutting historical data of the light guide film, the intelligence and self-learning ability of the system are further improved, the cutting process is closer to an ideal state, and due to the improvements, the production efficiency is improved, meanwhile, the material waste is reduced, and the production cost is reduced. And the cutting and positioning accuracy of the light guide film is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of light guide films, and in particular to a light guide film cutting and positioning system based on machine vision. Background Art

[0002] Light-guiding film technology involves the use of special films to evenly disperse light. It is mainly used in liquid crystal display (LCD) backlight sources and lighting systems. Light-guiding film is an optical-grade material that effectively controls the propagation path of light through specific pattern design and light management technology to achieve uniform light output. This material is commonly used in tablets, smart phones, TVs and other display devices. The key lies in how to design the film so that it can achieve high-efficiency light transmission and dispersion in the thinnest form possible, thereby improving the overall light efficiency and energy efficiency of the device.

[0003] Among them, the machine vision light-guiding film cutting and positioning system is a system that uses machine vision technology to optimize the light-guiding film production process. Machine vision is used to accurately identify the position and direction of the light-guiding film for precise cutting. Its purpose is to improve the degree of automation and accuracy of the cutting process, reduce material waste, and ensure that each piece of light-guiding film meets strict quality standards. Machine vision can achieve real-time adjustment of the cutting line and optimize production efficiency, making it suitable for high-output industrial environments.

[0004] Existing technologies show obvious limitations in high-efficiency and high-precision production environments, especially in high-volume industrial applications. Traditional technologies have limited capabilities in real-time data processing and dynamic adjustment, which often lead to insufficient cutting accuracy and material waste. They lack the ability to adjust tool position or pressure in real time. When processing light-guiding films with complex edge designs, multiple adjustments and repeated cutting are required, which increases production costs and time consumption. In addition, traditional systems have limited edge recognition and error correction capabilities and cannot effectively respond to rapidly changing production needs, resulting in inconsistent finished product quality and affecting the light efficiency and energy efficiency of display devices. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a light guide film cutting and positioning system based on machine vision.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a light guide film cutting and positioning system based on machine vision, the system comprising: The light guide film image processing module collects real-time image data of the light guide film, performs preliminary image format conversion, adjusts image clarity parameters to match processing requirements, and performs key frame screening to optimize image contrast and color saturation to obtain image quality enhancement data; The edge feature extraction module performs image edge detection based on the image quality enhancement data using a convolutional neural network, identifies edge features of the light guide film, extracts edge clarity and continuity features, calculates edge error values, and adjusts feature extraction thresholds to generate edge feature data sets; The tool adjustment module calculates the optimal angle and pressure of the cutting tool based on the edge characteristic data set, adjusts the position of the robot arm according to the real-time cutting feedback information, and dynamically optimizes the tool parameters, including repeatedly adjusting the tool angle and pressure to generate dynamic adjustment parameters for cutting; The cutting effect evaluation module dynamically adjusts the cutting parameters based on the cutting, monitors the current cutting effect, compares it with the preset cutting standard, executes the back propagation algorithm according to the deviation, optimizes the convolutional neural network in combination with the cutting history data of the light guide film, and generates a cutting effect optimization result.

[0007] The present invention is improved in that the steps of matching processing requirements are specifically as follows: Collecting real-time image data of the light-guiding film, performing preliminary format conversion of the image data, converting the original image data into a standard processing format, and obtaining formatted image data; The formatted image data is analyzed and image definition parameters are adjusted using the formula: ; Get the adjusted clarity , matching the image processing requirements, where is the original clarity, is the clarity adjustment factor, is the adjustment factor for the image baseline brightness.

[0008] The present invention is improved in that the step of acquiring the image quality enhancement data is specifically as follows: According to the motion characteristics and information content of the image, the key frame is screened to obtain key frame data; Based on the key frame data, the image quality is processed, including optimizing contrast and color saturation, using the formula: ; Get image quality enhancement data, where: To enhance the image quality, is the original image quality of the key frame, and They are contrast and saturation adjustment coefficients respectively.

[0009] The present invention is improved in that the step of identifying the edge features of the light guide film is specifically as follows: Based on the image quality enhancement data, using a convolutional neural network, setting network layer parameters that match the characteristics of the light guide film, including the number of network layers, filter size, and activation function parameters, to obtain network configuration data; Based on the network configuration data, an image edge detection operation is performed using the formula: ; Identify the edge position of the light guide film in the image and obtain the edge detection result, where: is the position in the image The edge detection result is is the image at position along The gradient of the direction, is the image at position along Directional gradient; The edge detection result is analyzed to extract key edge features, including edge clarity and continuity features, to obtain the edge shape of the light guide film.

[0010] The present invention is improved in that the step of acquiring the edge characteristic data set is specifically as follows: The edge error value is calculated using the formula: ; Get the total edge error value ,in, is the number of edge points, It is The detected edge point positions, It is The expected edge point locations, It is The absolute difference between the position of an edge point and the expected position; Based on the total edge error value, the feature extraction threshold is adjusted to reduce the edge error, using the formula: ; Get a new feature extraction threshold ,in, is the original feature extraction threshold, is the adjustment factor, is the edge error value; The edge features are re-evaluated according to the new feature extraction threshold to obtain an edge feature data set.

[0011] The present invention is improved in that the step of adjusting the position of the mechanical arm is specifically as follows: Based on the edge characteristic data set, the optimal angle and pressure of the cutting tool are calculated, and sensor data in the real-time cutting process, including the current position, speed, cutting angle and pressure of the robot arm, are collected to form a cutting feedback data set; The cropping feedback dataset is analyzed using the formula: ; Calculate the new position of the robot arm ,in, is the current location, is the expected distance between the target position and the current position, is the difference between the measured position and the current position, is the adjustment factor; The new position of the robot arm is fed back to the control system, the operating parameters of the robot arm are updated in real time, the alignment and motion control during the cutting process are optimized, and the robot arm position optimization result is obtained.

[0012] The present invention is improved in that the step of obtaining the dynamic adjustment parameters of cutting is specifically as follows: Collect sensor data from the real-time cutting process, integrate the data, and build a cutting process data set; Based on the cutting process data set, the optimal tool angle and pressure required for each cutting are predicted, and according to the historical cutting effect and the feedback information of the current light guide film, the tool parameters are dynamically optimized to obtain the optimized tool parameters; The optimized tool parameters are applied to the cutting process, and the tool angle and pressure are repeatedly adjusted through the control system to obtain the dynamic adjustment parameters for cutting.

[0013] The present invention is improved in that the step of obtaining the cutting effect optimization result is specifically as follows: Based on the cutting dynamic adjustment parameters, real-time monitoring data of the cutting operation is collected, including tool speed, accuracy and motion trajectory, to form a real-time cutting monitoring data set; Based on the real-time cutting monitoring data set, compared with the preset cutting standard, the formula is used: ; Calculate the overall cutting error value ,in, It is The current measured value, It is The preset standard values ​​of the parameters, It is The weight factor of the parameter, is the total number of cutting parameters; Based on the overall cutting error value, a back propagation algorithm is executed, and the convolutional neural network parameters are adjusted in combination with the historical cutting data of the light-guiding film to obtain a cutting effect optimization result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the accuracy and efficiency of light-guiding film cutting are significantly improved through real-time image data acquisition and edge detection and feature extraction of convolutional neural networks, allowing the system to accurately identify the edge position of the light-guiding film, reduce errors and optimize the cutting quality. The real-time feedback mechanism allows the system to make immediate adjustments based on the cutting effect, thereby improving the adaptability and continuity of the cutting operation. The convolutional neural network is optimized in combination with the cutting history data of the light-guiding film, further enhancing the intelligence and self-learning ability of the system, making the cutting process closer to the ideal state. These improvements not only improve production efficiency, but also reduce material waste, thereby ensuring the cutting and positioning accuracy of the light-guiding film. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a system flow chart of the present invention; Figure 2 A flowchart of matching processing requirements in the present invention; Figure 3 This is a flow chart for obtaining image quality enhancement data in the present invention; Figure 4 is a flow chart for identifying edge features of the light guide film in the present invention; Figure 5 A flowchart for obtaining edge characteristic data sets in the present invention; Figure 6 A flow chart of adjusting the position of the robot arm in the present invention; Figure 7 A flowchart of obtaining the dynamic adjustment parameters of the cutting in the present invention; Figure 8 This is a flow chart for obtaining the cutting effect optimization result in the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating positions or positional relationships, are based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0018] Example: See Figure 1 The present invention provides a technical solution: a light guide film cutting and positioning system based on machine vision comprises: The light guide film image processing module collects real-time image data of the light guide film, performs preliminary image format conversion, adjusts image clarity parameters to match processing requirements, and performs key frame screening to optimize image contrast and color saturation to obtain image quality enhancement data; The edge feature extraction module uses a convolutional neural network to perform image edge detection based on the image quality enhancement data, identify the edge features of the light guide film, extract the edge clarity and continuity features, calculate the edge error value, and adjust the feature extraction threshold to generate an edge feature data set; The tool adjustment module calculates the optimal angle and pressure of the cutting tool based on the edge characteristic data set, adjusts the position of the robot arm according to the real-time cutting feedback information, and dynamically optimizes the tool parameters, including repeatedly adjusting the tool angle and pressure to generate dynamic adjustment parameters for cutting; The cutting effect evaluation module dynamically adjusts the cutting parameters based on the cutting, monitors the current cutting effect, compares it with the preset cutting standard, executes the back propagation algorithm according to the deviation, optimizes the convolutional neural network based on the cutting history data of the light guide film, and generates the cutting effect optimization result.

[0019] Image quality enhancement data include resolution improvement results, dynamic range increase results, and signal-to-noise ratio improvement results. The edge characteristic data set specifically includes edge positioning accuracy, edge smoothness index, and edge contrast intensity. The dynamic adjustment parameters of cutting include angle fine-tuning accuracy, pressure adjustment sensitivity, and position calibration accuracy. The cutting effect optimization results specifically refer to the cutting accuracy improvement rate, error range reduction rate, and response time optimization information.

[0020] See also Figure 2 , the steps for matching processing requirements are as follows: Collecting real-time image data of the light-guiding film, performing preliminary format conversion of the image data, converting the original image data into a standard processing format, and obtaining formatted image data; By using a camera to capture the surface details of the light-guiding film, high-definition image acquisition is achieved. The captured image data is in raw data format before formatting. The raw data needs to be converted to meet the needs of subsequent processing. During the formatting process, the image data is compressed and transcoded and converted into a standard processing format such as JPEG or PNG. The conversion not only optimizes the data storage efficiency, but also facilitates subsequent image analysis and processing. The formatted image data is the converted standard image, which will be used for further image clarity adjustment and quality analysis.

[0021] Analyze the formatted image data and adjust the image clarity parameters using the formula: ; Get the adjusted clarity , matching the image processing requirements, where is the original clarity, is the clarity adjustment factor, which enhances the visibility of image details. It is the adjustment factor of the image baseline brightness, which is used to improve the overall visual effect of the image; By adjusting the parameters and , the image clarity can be flexibly adjusted according to different image processing requirements, thereby improving image quality and adapting to different monitoring and analysis conditions. 0.5, clarity adjustment coefficient 2.0, the adjustment factor of the image baseline brightness is 0.1, then the adjusted clarity The calculation is as follows: ; The results show that through appropriate parameter adjustment, the clarity is improved from 0.5 to 1.1, which significantly enhances the visibility of image details and adapts to more complex image processing requirements.

[0022] See also Figure 3 , the steps for obtaining image quality enhancement data are as follows: According to the motion characteristics and information content of the image, key frames are screened to obtain key frame data; A key frame screening process is designed. The displacement and change degree between frames are obtained by calculating the motion vector of each frame, and the motion characteristics of the frame are extracted. The content richness of each frame is analyzed in combination with the uniformity of pixel distribution and local gradient changes in the frame. This is used as an indicator to measure the information content in the frame. The motion feature value and content richness value are weighted to determine the importance of the frame. Frames with obvious motion characteristics and high content richness are screened as key frames. The key frame data are stored by numbering and indexing to form structured data that is convenient for subsequent calls.

[0023] Based on the key frame data, the image quality is processed, including optimizing contrast and color saturation, using the formula: ; Get image quality enhancement data, where: To enhance the image quality, is the original image quality of the key frame, and are contrast and saturation adjustment coefficients respectively; The following data was collected: the original image quality of the key frames is 0.6, the contrast adjustment factor 1.5, saturation adjustment factor is 0.05, and the calculation is: ; The results show that by properly adjusting the coefficient, the quality of the image is significantly improved from 0.6 to 0.95, indicating that the image is visually clearer and more vivid, suitable for high-quality display and analysis.

[0024] See also Figure 4 , the specific steps for identifying the edge features of the light guide film are: Based on the image quality enhancement data, the network layer parameters matching the characteristics of the light guide film are set using a convolutional neural network, including the number of network layers, filter size, and activation function parameters, to obtain network configuration data; Based on the image quality enhancement data, the configuration of the convolutional neural network layer parameters is adjusted to match the characteristics of the light-guiding film, including optimizing the number of network layers to adapt to the extraction of high-dimensional image features. The filter size needs to be adjusted according to the image characteristics to ensure the retention of details and the prominence of features. Specifically, the appropriate kernel size (such as 3×3 or 5×5 kernels) is selected for feature extraction according to the image resolution. The activation function parameters need to select parameter values ​​that can perform well in nonlinear feature learning, such as ReLU (linear rectification function) to ensure that gradient vanishing is avoided in feature learning. After the optimal parameter configuration is verified through experiments, the network configuration data is generated. The data provides efficient and reliable basic support for edge detection tasks.

[0025] Based on the network configuration data, perform image edge detection operation using the formula: ; Identify the edge position of the light guide film in the image and obtain the edge detection result, where: is the position in the image The edge detection result is is the image at position along The gradient of the direction, is the image at position along Directional gradient, used to obtain the total edge strength from the combination of gradients in the x and y directions; By synthesizing the gradients of the image in the x and y directions, the accuracy of edge recognition is effectively enhanced, making the edges more obvious, which is helpful for subsequent feature analysis and image processing tasks. The following data is collected. The gradient of the image in the x direction at point (x, y) is 150, and the gradient in the y direction is 100, then: ; The result shows that the edge strength at point (x, y) is 180.28, indicating that the point has a high edge characteristic and is a significant edge point in the image.

[0026] Analyze the edge detection results, extract key edge features, including edge clarity and continuity features, and obtain the edge shape of the light guide film; Analyze the edge detection results, and extract the clarity and continuity features of the edge of the light-guiding film to describe the edge shape. The specific steps include calculating the gradient amplitude of each edge point to measure the clarity, and using a standardized method to compare the clarity levels based on the gradient calculation results to ensure consistency. Continuity feature extraction determines the connectivity of the edge by calculating the Euclidean distance between adjacent points. Point pairs with too large a distance will be marked as breakpoints and repaired through an interpolation algorithm. The final processing result includes the clarity value of each edge point and the overall connectivity information of the edge line, generating a complete edge shape of the light-guiding film.

[0027] See also Figure 5 ,The specific steps for obtaining the edge feature dataset are: Calculate the edge error value using the formula: ; Get the total edge error value ,in, is the number of edge points, It is The detected edge point positions, It is The expected edge point locations, It is The absolute difference between the position of an edge point and the expected position; By quantifying the difference between the actual detected edge and the preset ideal edge, a method for directly measuring the error is provided, so that the accuracy of the edge detection algorithm can be quantitatively evaluated. When there are 3 edge points, the expected edge point positions are 10, 20 and 30 respectively, and the detected edge positions are 12, 18 and 33 respectively, substitute into the formula for calculation: ; ; The result shows that the total edge error value is 7, which indicates that there is a small deviation between the detection system and the ideal model, indicating that the edge detection needs to be fine-tuned to reduce the error.

[0028] Based on the total edge error value, adjust the feature extraction threshold to reduce the edge error using the formula: ; Get a new feature extraction threshold ,in, is the original feature extraction threshold, is the adjustment factor, which is used to adjust the threshold value according to the edge error. is the edge error value; Based on the calculated total edge error value 7, the feature extraction threshold is adjusted by the formula. If the original threshold is 50, adjustment factor is 0.1, and the new threshold is calculated by the formula: ; The adjustment reduces edge errors and improves edge detection accuracy, and the new threshold will be used in further image processing to ensure higher quality edge identification.

[0029] Re-evaluate edge features according to the new feature extraction threshold to obtain an edge feature dataset; The edge features are re-evaluated according to the adjusted feature extraction threshold. First, the edge detection results are matched and compared with the new threshold, and the edge information that meets the new threshold standard is extracted and marked as a valid edge. The marked edge information is further analyzed by calculating the clarity and continuity indicators of each edge to ensure the integrity and consistency of the edge features. During the analysis process, the edges are grouped based on the edge clarity value, and the position of the edge points of each group and their mutual connection methods are recorded. The length, direction and continuity of the edges are evaluated one by one. At the same time, the edge points that do not meet the clarity standard are removed, and the edge connection relationship is regenerated. Finally, the recalculated and optimized edge features are organized into a complete edge feature data set.

[0030] See also Figure 6 , the steps to adjust the position of the robot arm are as follows: Based on the edge characteristic data set, the optimal angle and pressure of the cutting tool are calculated, and the sensor data in the real-time cutting process is collected, including the current position, speed, cutting angle and pressure of the robot arm, to form a cutting feedback data set; Based on the edge characteristic data set, we first call the physical characteristic parameters such as edge length, thickness and material hardness recorded in the data set, and normalize the parameters respectively to calculate the theoretical range of cutting pressure and cutting angle, and calculate the expected value of cutting angle through the normalized edge length. The required pressure range for cutting is calculated through the normalized material hardness combined with the edge thickness. At the same time, a weight coefficient is added after normalization to adjust the dynamic sensitivity of the actual cutting pressure. Finally, the optimal angle and pressure are selected as cutting parameters for actual cutting tasks according to the calculation results to form a cutting feedback data set.

[0031] Analyze the cutting feedback dataset and use the formula: ; Calculate the new position of the robot arm ,in, is the current location, is the expected difference between the target position and the current position, and is the expected difference between the target position and the current position set in the optimal cutting path planning. It is the difference between the measured position and the current position, and is the actual difference between the current position and the theoretical target position measured during the actual cutting process. is the adjustment coefficient, which is used to adjust the position of the robot according to the position error; By adjusting the position of the robot arm in real time and dynamically responding to deviations in the cutting process, the cutting accuracy and mechanical response speed are enhanced, so that the robot arm can more accurately align with the target cutting position. is 5.0, target position deviation is 0.5, the current measured position deviation is 0.3, adjustment coefficient Set it to 1.2 and calculate according to the formula: ; The results show that the new robot arm position is adjusted to 5.24 to adapt to the actual cutting requirements and reduce cutting errors.

[0032] Feedback the new position of the robot arm to the control system (the control system is mainly used to control the robot arm and the tool), update the operating parameters of the robot arm in real time, optimize the alignment and motion control during the cutting process, and obtain the optimization result of the robot arm position; By calling the real-time position, pressure and speed recorded in the sensor data, the motion trajectory of the robot is optimized in real time in combination with the new position parameters calculated by feedback. The current position of the robot and the calculated new position are called for path planning respectively to ensure that the speed changes smoothly and without errors during the path adjustment process. At the same time, the correlation between the cutting pressure change range and the actual pressure adjustment is calculated based on the real-time feedback, and pressure adjustment compensation is performed on the motion control. Finally, the optimized cutting motion parameters are output and fed back to the control system to update the operation instructions in real time to obtain the robot position optimization result.

[0033] See also Figure 7 , the specific steps for obtaining the dynamic adjustment parameters of cutting are: Collect sensor data from the real-time cutting process, integrate the data, and build a cutting process data set; The extracted sensor data includes the current position, cutting pressure and speed of the robot arm. By aligning the data with a unified timestamp and standardizing the data format, the data at each moment in the cutting process are integrated into a complete cutting process data set. At the same time, abnormal data that obviously deviates from the cutting range is eliminated, and the robot arm's movement path and actual cutting force at each moment are marked as a separate parameter set.

[0034] Based on the cutting process data set, the optimal tool angle and pressure required for each cutting are predicted, and according to the historical cutting effect and the feedback information of the current light guide film, the tool parameters are dynamically optimized to obtain the optimized tool parameters; Based on the cutting process data set, by calling the actual cutting effect and material feedback parameters in the historical cutting data, the parameters are correlated with the cutting force and cutting path in the current data. Through the dynamic calculation model, the difference between the actual cutting path and the ideal cutting trajectory is matched. According to the cutting response characteristics of different materials, the optimization range of the tool angle and the adjustment range of the pressure are calculated one by one. Finally, the optimization range is reduced according to the cutting accuracy requirements, and the optimized tool parameters are finally obtained.

[0035] The optimized tool parameters are applied to the cutting process, and the tool angle and pressure are repeatedly adjusted through the control system to obtain the dynamic adjustment parameters of cutting; The optimized tool parameters are applied to the cutting process. The optimized cutting parameters are loaded through the real-time control system, the motion trajectory of the robot arm is dynamically adjusted, and the fluctuation of the cutting pressure is monitored in real time. The tool angle adjustment amplitude at each moment is compared with the cutting effect fed back in real time. When the pressure or cutting angle deviates from the optimization range, the corresponding parameters are adjusted immediately. The process is repeated until the cutting is completed to obtain the dynamic adjustment parameters for cutting.

[0036] See also Figure 8 , the specific steps for obtaining the cutting effect optimization results are: Based on the dynamic adjustment parameters of cutting, real-time monitoring data of cutting operations are collected, including tool speed, accuracy and motion trajectory, to form a real-time cutting monitoring data set; Based on the framework of dynamic adjustment parameters of cutting, cutting operation data is collected in real time. The speed, accuracy and motion trajectory information of the tool are captured through sensors. The collected raw data is time-stamped and denoised, and abnormal data is eliminated. The data is reintegrated into a unified data structure according to the time series. At the same time, the cutting position and trajectory deviation corresponding to each moment are recorded, and the data is classified and grouped according to the cutting parameters to form a real-time monitoring data set for the cutting operation. The data provides a basis for subsequent cutting quality evaluation and adjustment.

[0037] Based on the real-time cutting monitoring data set, compared with the preset cutting standard, the formula is used: ; Calculate the overall cutting error value ,in, It is The current measured value, that is, the cutting parameter value actually recorded during the cutting process, It is The preset standard value of each parameter, that is, the ideal or standard parameter value set before the cutting operation, It is The weight factor of each parameter indicates the importance of each parameter in the error calculation. is the total number of cutting parameters; There are three parameters, 50, 55 and 60 respectively. They are 48, 53 and 58 respectively. The weight factors are 0.5, 1.0, and 1.5 respectively, and are substituted into the formula for calculation: ; ; The result shows that the calculated overall cutting error value is 11.0, which indicates the degree of deviation between the actual operation and the preset standard, and thus provides a quantitative basis for the adjustment of cutting parameters.

[0038] Based on the overall cutting error value, the back propagation algorithm is executed, and the convolutional neural network parameters are adjusted in combination with the historical cutting data of the light guide film to obtain the cutting effect optimization result; Based on the overall cutting error value, the real-time feedback in the cutting process is compared with the preset cutting standard, each set of error data is clustered and analyzed, and the parameters with large cutting deviations are taken as optimization targets. The convolutional neural network is trained using the historical cutting data of the light-guiding film, and the weights and bias values ​​in the neural network are adjusted through back propagation to ensure that the calculation of the cutting parameters is highly consistent with the current feedback. After re-iteration optimization, the cutting effect optimization result is generated, which plays a dynamic adjustment role in subsequent cutting operations.

[0039] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A light guide film cutting and positioning system based on machine vision, characterized in that: The system comprises: The light guide film image processing module collects real-time image data of the light guide film, performs preliminary image format conversion, adjusts image clarity parameters to match processing requirements, and performs key frame screening to optimize image contrast and color saturation to obtain image quality enhancement data; The edge feature extraction module performs image edge detection based on the image quality enhancement data using a convolutional neural network, identifies edge features of the light guide film, extracts edge clarity and continuity features, calculates edge error values, and adjusts feature extraction thresholds to generate edge feature data sets; The tool adjustment module calculates the optimal angle and pressure of the cutting tool based on the edge characteristic data set, adjusts the position of the robot arm according to the real-time cutting feedback information, and dynamically optimizes the tool parameters, including repeatedly adjusting the tool angle and pressure to generate dynamic adjustment parameters for cutting; The cutting effect evaluation module dynamically adjusts the cutting parameters based on the cutting, monitors the current cutting effect, compares it with the preset cutting standard, executes the back propagation algorithm according to the deviation, optimizes the convolutional neural network in combination with the cutting history data of the light guide film, and generates a cutting effect optimization result.

2. The light guide film cutting and positioning system based on machine vision according to claim 1, characterized in that: The steps of matching processing requirements are specifically as follows: Collecting real-time image data of the light-guiding film, performing preliminary format conversion of the image data, converting the original image data into a standard processing format, and obtaining formatted image data; The formatted image data is analyzed and image definition parameters are adjusted using the formula: ； Get the adjusted clarity , matching the image processing requirements, where is the original clarity, is the clarity adjustment factor, is the adjustment factor for the image baseline brightness.

3. The machine vision-based light guide film cutting and positioning system according to claim 1, characterized in that: The steps of acquiring the image quality enhancement data are specifically as follows: According to the motion characteristics and information content of the image, the key frame is screened to obtain key frame data; Based on the key frame data, the image quality is processed, including optimizing contrast and color saturation, using the formula: ； Get image quality enhancement data, where: To enhance the image quality, is the original image quality of the key frame, and They are contrast and saturation adjustment coefficients respectively.

4. The machine vision-based light guide film cutting and positioning system according to claim 1, characterized in that: The steps for identifying the edge features of the light guide film are specifically as follows: Based on the image quality enhancement data, using a convolutional neural network, setting network layer parameters that match the characteristics of the light guide film, including the number of network layers, filter size, and activation function parameters, to obtain network configuration data; Based on the network configuration data, an image edge detection operation is performed using the formula: ； Identify the edge position of the light guide film in the image and obtain the edge detection result, where: is the position in the image The edge detection result is is the image at position along The gradient of the direction, is the image at position along Directional gradient; The edge detection result is analyzed to extract key edge features, including edge clarity and continuity features, to obtain the edge shape of the light guide film.

5. The light guide film cutting and positioning system based on machine vision according to claim 1, characterized in that: The steps for obtaining the edge feature data set are specifically as follows: The edge error value is calculated using the formula: ； Get the total edge error value ,in, is the number of edge points, It is The detected edge point positions, It is The expected edge point locations, It is The absolute difference between the position of an edge point and the expected position; Based on the total edge error value, the feature extraction threshold is adjusted to reduce the edge error, using the formula: ； Get a new feature extraction threshold ,in, is the original feature extraction threshold, is the adjustment factor, is the edge error value; The edge features are re-evaluated according to the new feature extraction threshold to obtain an edge feature data set.

6. The light guide film cutting and positioning system based on machine vision according to claim 1, characterized in that: The steps of adjusting the position of the robotic arm are specifically as follows: Based on the edge characteristic data set, the optimal angle and pressure of the cutting tool are calculated, and sensor data in the real-time cutting process, including the current position, speed, cutting angle and pressure of the robot arm, are collected to form a cutting feedback data set; The cropping feedback dataset is analyzed using the formula: ； Calculate the new position of the robot arm ,in, is the current location, is the expected distance between the target position and the current position, is the difference between the measured position and the current position, is the adjustment factor; The new position of the robot arm is fed back to the control system, the operating parameters of the robot arm are updated in real time, the alignment and motion control during the cutting process are optimized, and the robot arm position optimization result is obtained.

7. The light guide film cutting and positioning system based on machine vision according to claim 1, characterized in that: The steps for obtaining the dynamic adjustment parameters of cutting are specifically as follows: Collect sensor data from the real-time cutting process, integrate the data, and build a cutting process data set; Based on the cutting process data set, the optimal tool angle and pressure required for each cutting are predicted, and according to the historical cutting effect and the feedback information of the current light guide film, the tool parameters are dynamically optimized to obtain the optimized tool parameters; The optimized tool parameters are applied to the cutting process, and the tool angle and pressure are repeatedly adjusted through the control system to obtain the dynamic adjustment parameters for cutting.

8. The machine vision-based light guide film cutting and positioning system according to claim 1, characterized in that: The steps for obtaining the cutting effect optimization result are specifically as follows: Based on the cutting dynamic adjustment parameters, real-time monitoring data of the cutting operation is collected, including tool speed, accuracy and motion trajectory, to form a real-time cutting monitoring data set; Based on the real-time cutting monitoring data set, compared with the preset cutting standard, the formula is used: ； Calculate the overall cutting error value ,in, It is The current measured value, It is The preset standard values ​​of the parameters, It is The weight factor of the parameter, is the total number of cutting parameters; Based on the overall cutting error value, a back propagation algorithm is executed, and the convolutional neural network parameters are adjusted in combination with the historical cutting data of the light-guiding film to obtain a cutting effect optimization result.

Images