Pattern texture offset detection method and system based on dual-camera different-view-angle imaging

Abstract

The invention discloses a pattern texture offset detection method and system based on dual-camera different-view-angle imaging, and relates to the technical field of pattern texture offset detection, and the method comprises the following steps: S1, camera installation: selecting two black and white cameras with the same resolution, and respectively carrying out vertical installation and low-angle inclined installation on the two black and white cameras; s2, template benchmark acquisition: when a product pattern and a texture are in a standard alignment state; according to the pattern texture offset detection method and system based on double-camera different-view-angle imaging, the double-camera design is adopted, compared with a single-camera scheme, high-quality pattern information and texture information can be obtained at the same time, the stability and accuracy of offset detection can be remarkably improved, and compared with a scheme needing color codes and positioning holes, the method and system have the advantages that the cost is reduced, and the efficiency is improved. According to the method, additional marks are not needed, the application range is wider, the modification cost is lower, meanwhile, through hardware constraints of the same resolution ratio and the same object distance, the deviator can be directly calculated through the pixel difference, and calibration and parameter maintenance are simplified.

Description

Technical Field

[0001] This invention relates to the field of pattern and texture misalignment detection technology, specifically to a pattern and texture misalignment detection method and system based on dual-camera different-view imaging. Background Technology

[0002] In continuous production processes such as flooring decorative paper lamination, board bonding printing, film lamination, and textile printing lamination, there is often an overlap relationship between the pattern layer and the substrate texture layer / base layer. During production, factors such as unwinding tension fluctuations, guide roller wear, correction response delays, material expansion and contraction, hot pressing, or adhesive layer flow can easily lead to misalignment (registration misalignment) between the two layers. This manifests as a lateral / vertical shift in the pattern relative to the texture, causing quality problems such as pattern misalignment, exposed edges, and uneven pattern matching. To ensure product appearance consistency and yield, the production line usually needs to perform online detection of the relative position of the pattern and texture, and trigger alarms or linkage correction / rejection when the deviation exceeds the tolerance. However, existing detection methods often use a single-camera, single-viewpoint approach, which makes it difficult to balance the imaging quality of patterns and textures. Pattern information is usually best acquired at a near-vertical viewpoint to reduce perspective distortion and improve geometric consistency. Texture information may have insufficient contrast or severe reflection under vertical lighting / viewpoint, often requiring a low-angle viewpoint or side lighting to enhance texture undulations and details. Therefore, single-viewpoint imaging often results in "clear patterns but weak textures / clear textures but distorted patterns," affecting the stability and accuracy of offset measurement. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for detecting pattern and texture misalignment based on dual-camera different-view imaging, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method and system for detecting pattern texture misalignment based on dual-camera heterogeneous view imaging, comprising the following steps: S1. Camera Installation: Select two monochrome cameras with the same resolution, and install them vertically and at a low angle, respectively. S2, Template Reference Acquisition: When the product pattern and texture are in a standard alignment state, that is, a qualified reference state with no misalignment or deviation, two cameras are simultaneously triggered to complete the reference image acquisition. S3, ROI Configuration and Template Establishment: Based on two template images acquired by two cameras, the detection area and baseline parameters are calibrated, and a one-to-one detection matching relationship is established; S4. Online acquisition and preprocessing: During normal production line operation, real-time image acquisition and preprocessing are performed synchronously to provide high-quality image data for subsequent positioning calculations; S5. Dual-channel positioning and displacement calculation: For each ROI, template matching positioning is performed in the pattern image and texture image to obtain the positioning coordinates. Then, the relative displacement between the two positioning coordinates is calculated, and reference compensation is introduced to obtain the error. S6. Partition Threshold Determination and Result Output: Independently determine the bias error of each ROI, and optionally fuse the results of multiple ROIs to output the final detection conclusion and related data to support subsequent processing on the production line.

[0005] Furthermore, in step S1, the first camera uses an approximately vertical viewing angle to image the pattern on the product surface, while the second camera uses a low-angle viewing angle to image the texture on the product surface. The object distance between the two cameras must be the same, and the two cameras must be aligned with the same line position when installed on the production line so that the relative displacement calculation can be performed under the same coordinate reference.

[0006] Furthermore, in step S2, the first camera is used to acquire pattern template images, and the second camera is used to acquire texture template images. The two template images have completely consistent pixel scales in the horizontal direction, providing a unified pixel benchmark for subsequent ROI matching and displacement calculation.

[0007] Furthermore, step S3 includes the following sub-steps: S31. Multiple ROI Area Settings: In the pattern template image and texture template image, multiple detection ROI areas with the same position and size are set, and each ROI corresponds to a local area on the product that needs to be detected to avoid misalignment, ensuring that the detection areas of the two channels correspond one to one. S32. Feature Block Extraction: For each corresponding ROI region, extract pattern feature blocks (such as unique features like pattern outline, corners, and logos) and texture feature blocks (such as unique features like texture, particle distribution, and light and dark textures) as core references for subsequent online template matching. S33, Threshold and Compensation Recording: Record two sets of key parameters independently for each ROI region.

[0008] Furthermore, in step S33, the allowable deviation threshold (X / Y direction) is the maximum allowable deviation error in the ROI area. If the deviation exceeds this threshold, it is determined to be a local deviation anomaly. The reference compensation amount (X / Y direction) is a compensation parameter used to eliminate system deviations such as minor installation errors of the dual cameras and uneven illumination, thereby improving the accuracy of positioning calculation.

[0009] Furthermore, step S4 includes the following sub-steps: S41. Synchronous online acquisition: Real-time and synchronous triggering of dual cameras to automatically acquire the pattern and texture images of the product under test. S42. Preprocessing: Perform a unified preprocessing operation on the two images to be tested to automatically eliminate environmental interference and improve feature recognizability. The processing flow includes: Fixed interference area masking: Pre-mark fixed interference areas such as production line background, fixtures, and stains, and perform masking to avoid the influence of interference features on the matching results; Scale normalization or downsampling: According to the actual detection efficiency and accuracy requirements, the image is scale normalized (to ensure that the scale is completely consistent with the template image) or reasonably downsampled (to reduce the amount of calculation and improve the detection frame rate).

[0010] Furthermore, the formula for calculating the relative displacement in step S5 is as follows: (in , : No. The actual location coordinates of each ROI in the pattern image to be tested; , : No. The actual location coordinates of each ROI in the textured image to be tested; , : No. (Original relative displacement of each ROI); Formula for calculating the final error after benchmark compensation: (in , : No. The baseline compensation amount for each ROI; , : No. (final bias error of each ROI after compensation), if >0: This indicates that the texture is offset to the right horizontally relative to the pattern. <0: shift to the left; if >0: This indicates that the texture is offset vertically upwards relative to the pattern. <0: Offset downwards.

[0011] Furthermore, the aforementioned , The calibration logic is as follows: In step S3, the template image ROI under standard alignment is located in two channels, and the original relative displacement at this time is calculated. The negative number is the reference compensation amount (used to offset the inherent deviation of the system).

[0012] A pattern and texture misalignment detection system based on dual-camera anisotropic imaging, wherein the misalignment detection system is applied to the aforementioned misalignment detection method, and the pattern and texture misalignment detection system based on dual-camera anisotropic imaging includes: Camera video control module: used to automatically control the installed camera to perform intermittent video recording of the conveyed products, and send the information data collected by the camera to subsequent modules for calculation and processing; Displacement numerical calculation module: It is used to automatically receive information data collected by the camera control module, input the received information data into the model for displacement calculation, and then send the calculated data to the subsequent modules for judgment and processing; The regional value determination module is used to independently determine the calculated deviation error and automatically output the final detection conclusion and related data to support subsequent processing on the production line.

[0013] Furthermore, the camera shooting control module includes a camera shooting control unit, a template reference acquisition unit, an acquisition data preprocessing unit, and a data output unit. The camera shooting control unit is used to automatically control the intermittent operation and shutdown of the two cameras. The template reference acquisition unit is used to receive reference image information data acquired by the two cameras. The acquisition data preprocessing unit is used to preprocess the acquired data to remove invalid data such as blurred images.

[0014] This invention provides a method and system for detecting pattern and texture misalignment based on dual-camera different-view imaging, which has the following advantages: 1. This invention adopts a dual-camera design, which can simultaneously obtain high-quality pattern information and texture information compared with the single-camera solution, significantly improving the stability and accuracy of offset detection. Moreover, compared with the solution that requires color marks and positioning holes, this invention does not require additional markings, has a wider range of applications, and lower modification costs. At the same time, through hardware constraints of the same resolution and the same object distance, the offset can be directly calculated from the pixel difference, simplifying calibration and parameter maintenance.

[0015] 2. This invention employs a multi-ROI and partition threshold mechanism, which effectively improves robustness to complex patterns, local contamination, reflection and other working conditions, thereby reducing the probability of false alarms and missed alarms. It can also output deviation and abnormal location, thereby providing quantitative basis for subsequent rejection and closed-loop correction control, so that subsequent production can proceed normally. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall operation process of the pattern and texture offset detection method based on dual-camera different-view imaging according to the present invention. Figure 2 This is a schematic diagram of the operation process of a pattern and texture offset detection system based on dual-camera different-view imaging according to the present invention. Figure 3 This is a schematic diagram of the operation flow of the camera control module of a pattern and texture offset detection system based on dual-camera different-view imaging according to the present invention. Figure 4This is a schematic diagram of the camera installation position for a pattern and texture offset detection method and system based on dual-camera different-view imaging according to the present invention. Detailed Implementation

[0017] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0018] like Figures 1-4 As shown, a pattern texture offset detection method based on dual-camera different-view imaging includes the following steps: S1. Camera Installation: Select two black and white cameras with the same resolution and install them vertically and at a low angle, respectively. The first camera uses an approximately vertical viewing angle to image the pattern on the product surface, while the second camera uses a low angle viewing angle to image the texture on the product surface. The object distance between the two cameras must be the same, and the two cameras must be aligned with the same line position when installed on the production line so that the relative displacement calculation can be performed under the same coordinate reference. S2. Template reference acquisition: When the product pattern and texture are in a standard alignment state, that is, a qualified reference state with no misalignment or deviation, two cameras are simultaneously triggered to complete the reference image acquisition. The first camera is used to acquire the pattern template image, and the second camera is used to acquire the texture template image. The two template images are completely consistent in pixel scale in the horizontal direction, providing a unified pixel reference for subsequent ROI matching and displacement calculation. S3. ROI Configuration and Template Establishment: Based on two template images acquired by two cameras, the detection area and baseline parameters are calibrated, and a one-to-one detection matching relationship is established. This includes the following sub-steps: S31. Multiple ROI Area Settings: In the pattern template image and texture template image, multiple detection ROI areas with the same position and size are set, and each ROI corresponds to a local area on the product that needs to be detected to avoid misalignment, ensuring that the detection areas of the two channels correspond one to one. S32. Feature Block Extraction: For each corresponding ROI region, extract pattern feature blocks (such as unique features like pattern outline, corners, and logos) and texture feature blocks (such as unique features like texture, particle distribution, and light and dark textures) as core references for subsequent online template matching. S33. Threshold and Compensation Record: Two sets of key parameters are recorded independently for each ROI region. Allowable deviation threshold (X / Y direction): The maximum allowable deviation error for this ROI region. If the deviation exceeds this threshold, it is judged as a local deviation anomaly. Baseline compensation (X / Y direction): Compensation parameters used to eliminate system deviations such as minor errors in dual camera installation and uneven illumination, thereby improving the accuracy of positioning calculation. S4. Online Image Acquisition and Preprocessing: During normal production line operation, real-time image acquisition and preprocessing are performed synchronously to provide high-quality image data for subsequent positioning calculations. This includes the following sub-steps: S41. Synchronous online acquisition: Real-time and synchronous triggering of dual cameras to automatically acquire the pattern and texture images of the product under test. S42. Preprocessing: Perform uniform preprocessing on the two images to be tested to automatically eliminate environmental interference and improve feature recognizability. The processing flow includes: Fixed interference area masking: Pre-mark fixed interference areas such as production line background, fixtures, and stains, and perform masking to avoid the influence of interference features on the matching results; Scale normalization or downsampling: According to the actual detection efficiency and accuracy requirements, the image is scale normalized (to ensure that the scale is completely consistent with the template image) or reasonably downsampled (to reduce the amount of computation and improve the detection frame rate). S5. Dual-channel positioning and displacement calculation: For each ROI, template matching is performed in both the pattern and texture images to obtain the positioning coordinates. Then, the relative displacement between the two positioning coordinates is calculated, and a reference compensation is introduced to obtain the error. The formula for relative displacement calculation is as follows: (in , : No. The actual location coordinates of each ROI in the pattern image to be tested; , : No. The actual location coordinates of each ROI in the textured image to be tested; , : No. (Original relative displacement of each ROI); Formula for calculating the final error after benchmark compensation: (in , : No. The baseline compensation amount for each ROI; , : No. (final bias error of each ROI after compensation), if >0: This indicates that the texture is offset to the right horizontally relative to the pattern. <0: shift to the left; if >0: This indicates that the texture is offset vertically upwards relative to the pattern. <0: Downward offset, , The calibration logic is as follows: In step S3, the template image ROI under standard alignment is located in two channels, and the original relative displacement at this time is calculated. The negative number is the reference compensation amount (used to offset the inherent deviation of the system). S6. Partition Threshold Determination and Result Output: Independently determine the bias error of each ROI, and optionally fuse the results of multiple ROIs to output the final detection conclusion and related data to support subsequent processing on the production line.

[0019] like Figure 2 and Figure 3 As shown, a pattern texture misalignment detection system based on dual-camera different-view imaging is provided. This misalignment detection system is applied to the aforementioned misalignment detection method and includes: Camera control module: This module is used to automatically control the installed cameras to perform intermittent video recording of the conveyed products and send the acquired video data to subsequent modules for processing. The camera control module includes a camera shooting control unit, a template reference acquisition unit, an acquisition data preprocessing unit, and a data output unit. The camera shooting control unit is used to automatically control the intermittent operation and shutdown of the two cameras. The template reference acquisition unit is used to receive reference image information data acquired by the two cameras. The acquisition data preprocessing unit is used to preprocess the acquired data to remove invalid data such as blurry images. Displacement numerical calculation module: It is used to automatically receive information data collected by the camera control module, input the received information data into the model for displacement calculation, and then send the calculated data to the subsequent modules for judgment and processing; The regional value determination module is used to independently determine the calculated deviation error and automatically output the final detection conclusion and related data to support subsequent processing on the production line.

[0020] Example: S1. Camera Installation and Calibration: Fix the first camera directly above the production line using a vertical bracket, with the lens axis perpendicular to the product surface (perpendicularity deviation ≤ 0.5°), to capture the line patterns on the product surface; mount the second camera on the same side using a tilt bracket, adjust the tilt angle to 30°, with the lens facing the product surface, to capture the weave texture of the underlying substrate. Calibrate the object distance between the two cameras using a standard calibration plate (accuracy 0.001mm) to ensure that both are 400mm. Align the line positions using a crosshair cursor. After calibration, the inherent deviation of the system should be ≤ 0.005mm. S2. Template Reference Acquisition: Select 10 qualified products with standard alignment (confirmed by manual inspection that the pattern and texture are not misaligned, with a deviation ≤0.003mm), place them at the production line inspection station, trigger the template reference acquisition unit, and simultaneously acquire the pattern template image and texture template image of each product, obtaining 10 sets of images for each. Filter the clarity of each set of images (clarity evaluation index ≥0.8, calculated using the variance method), select the 3 best sets of images as the reference template library, and take the average value to reduce random errors; S3, ROI Configuration and Template Creation: S31. Multiple ROI Area Settings: Based on the baseline template image, manually mark 4-6 ROI areas in Halcon software, corresponding to key positions on the product. Set the size of each ROI to 80×80 pixels (corresponding to the actual size: 80×0.3mm=24mm) to ensure that the position and size of each ROI are completely consistent in the pattern template and texture template. S32. Feature Block Extraction: Feature block extraction is performed on each ROI region. For pattern ROI, the contour extraction algorithm (Canny edge detection, threshold range [50, 150]) is used to extract features such as pad edges and line contours, and generate feature vectors (128 dimensions). For texture ROI, the LBP texture feature extraction algorithm (neighborhood radius 3, sampling points 8) is used to extract the particle distribution and brightness variation features of the substrate braided texture, and generate 128-dimensional feature vectors. Then, the feature blocks of all ROIs are stored in the template library as a reference benchmark for online matching. S33. Threshold and Compensation Record: Allowable Deviation Threshold Setting: According to the product technical requirements, the allowable deviation threshold for each ROI is uniformly set to ±2mm in the X direction and ±2mm in the Y direction (corresponding pixel value: ±2mm / (0.3mm / pixel)≈±6.6 pixels, rounded to ±6 pixels). Baseline Compensation Calibration: Perform dual-channel positioning on each ROI of the 3 sets of standard template images, calculate the original relative displacement, take the average value and then take the opposite number as the baseline compensation amount. S4. Online data acquisition and preprocessing: S41. Synchronous online acquisition: When the production line is running, when the product arrives at the inspection station, the photoelectric sensor triggers the camera shooting control unit, which simultaneously triggers the dual cameras to acquire images. The acquisition cycle is 500ms (corresponding to a frame rate of 2fps, which is higher than the 1fps required by the production line). Each product acquires 1 frame of pattern image to be tested and 1 frame of texture image to be tested, which are transmitted to the preprocessing unit in real time. S42. Preprocessing: Fixed interference area masking: The built-in mask library is used to mask the fixture-occluded areas and production line background areas in the image, retaining only the effective product area. Scale normalization: Gray-level histogram normalization is performed (balancing the pixel value distribution of the overall image) to eliminate the influence of light source brightness fluctuations. Noise reduction: A Gaussian filtering algorithm is used (convolution kernel size 3×3, standard deviation...). =1.5), remove environmental noise (such as dust reflection, circuit interference), improve feature recognizability, and the preprocessing time is ≤5ms / frame; S5. Dual-channel positioning and displacement calculation: For the preprocessed image to be tested, the NCC matching algorithm is called for each ROI to match with the feature blocks in the template library, and the positioning coordinates of each ROI in the pattern image to be tested are obtained. , ) and the localization coordinates in the texture image to be tested ( , The matching success rate is ≥99.5%, the positioning accuracy is ±0.5 pixels, and the original relative displacement of each ROI is calculated according to the formula. , Then substitute the baseline compensation amounts for each ROI. , Calculate the final bias error , And convert it into the actual deviation value (actual deviation = pixel deviation × 0.3mm / pixel). S6. Regional Judgment and Result Output: The final deviation error of 4-6 ROIs is compared with the allowable threshold (±2mm) one by one. If the error of a single ROI exceeds the threshold, it is marked as a local deviation anomaly. If the errors of all ROIs are within the allowable range and the overall error is ≤±1.5mm, it is judged as qualified, and the test qualified conclusion and error data of each ROI are output. If there is one or more abnormal ROIs, or the overall error exceeds ±1.5mm, it is judged as unqualified, the local deviation anomaly conclusion is output, the abnormal ROI location and error value are marked, the production line alarm is triggered simultaneously (audio and visual alarm, alarm duration 3s), and the production line is controlled to divert defective products to the waste area.

[0021] The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and to design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A method for detecting pattern and texture misalignment based on dual-camera, different-view imaging, characterized in that, Includes the following steps: S1. Camera Installation: Select two monochrome cameras with the same resolution, and install them vertically and at a low angle, respectively. S2, Template Reference Acquisition: When the product pattern and texture are in a standard alignment state, that is, a qualified reference state with no misalignment or deviation, two cameras are simultaneously triggered to complete the reference image acquisition. S3, ROI Configuration and Template Establishment: Based on two template images acquired by two cameras, the detection area and baseline parameters are calibrated, and a one-to-one detection matching relationship is established; S4. Online acquisition and preprocessing: During normal production line operation, real-time image acquisition and preprocessing are performed synchronously to provide high-quality image data for subsequent positioning calculations; S5. Dual-channel positioning and displacement calculation: For each ROI, template matching positioning is performed in the pattern image and texture image to obtain the positioning coordinates. Then, the relative displacement between the two positioning coordinates is calculated, and reference compensation is introduced to obtain the error. S6. Partition Threshold Determination and Result Output: Independently determine the bias error of each ROI, and optionally fuse the results of multiple ROIs to output the final detection conclusion and related data to support subsequent processing on the production line.

2. The method for detecting pattern and texture misalignment based on dual-camera heterogeneous view imaging according to claim 1, characterized in that, In step S1, the first camera uses an approximately vertical viewing angle to image the pattern on the product surface, while the second camera uses a low-angle viewing angle to image the texture on the product surface. The object distance between the two cameras must be the same, and the two cameras must be aligned with the same line position when installed on the production line so that the relative displacement calculation can be performed under the same coordinate reference.

3. The pattern and texture misalignment detection method based on dual-camera heterogeneous view imaging according to claim 1, characterized in that, In step S2, the first camera is used to acquire pattern template images, and the second camera is used to acquire texture template images. The two template images have completely consistent pixel scales in the horizontal direction, providing a unified pixel benchmark for subsequent ROI matching and displacement calculation.

4. The pattern and texture misalignment detection method based on dual-camera different-view imaging according to claim 1, characterized in that, Step S3 includes the following sub-steps: S31. Multiple ROI Area Settings: In the pattern template image and texture template image, multiple detection ROI areas with the same position and size are set, and each ROI corresponds to a local area on the product that needs to be detected to avoid misalignment, ensuring that the detection areas of the two channels correspond one to one. S32. Feature block extraction: For each corresponding ROI region, extract pattern feature blocks and texture feature blocks respectively, which will serve as the core reference for subsequent online template matching; S33, Threshold and Compensation Recording: Record two sets of key parameters independently for each ROI region.

5. The pattern and texture misalignment detection method based on dual-camera different-view imaging according to claim 4, characterized in that, In step S33, the allowable deviation threshold is the maximum allowable deviation error in the ROI area. If the deviation exceeds this threshold, it is determined to be a local deviation anomaly. The reference compensation amount is a compensation parameter used to eliminate system deviations such as minor installation errors of the dual cameras and uneven illumination, thereby improving the accuracy of positioning calculation.

6. The pattern and texture misalignment detection method based on dual-camera different-view imaging according to claim 1, characterized in that, Step S4 includes the following sub-steps: S41. Synchronous online acquisition: Real-time and synchronous triggering of dual cameras to automatically acquire the pattern and texture images of the product under test. S42. Preprocessing: Perform a unified preprocessing operation on the two images to be tested to automatically eliminate environmental interference and improve feature recognizability. The processing flow includes: Fixed interference area masking: Pre-mark fixed interference areas such as production line background, fixtures, and stains, and perform masking to avoid the influence of interference features on the matching results. Scale normalization or downsampling: Based on the actual detection efficiency and accuracy requirements, the image is scale normalized or reasonably downsampled.

7. The pattern and texture misalignment detection method based on dual-camera different-view imaging according to claim 1, characterized in that, The formula for calculating relative displacement in step S5 is as follows: (in , : No. The actual location coordinates of each ROI in the pattern image to be tested; , : No. The actual location coordinates of each ROI in the textured image to be tested; , : No. (Original relative displacement of each ROI); Formula for calculating the final error after benchmark compensation: (in , : No. The baseline compensation amount for each ROI; , : No. (final bias error of each ROI after compensation), if >0: This indicates that the texture is offset to the right horizontally relative to the pattern. <0: shift to the left; if >0: This indicates that the texture is offset vertically upwards relative to the pattern. <0: Offset downwards.

8. The pattern and texture misalignment detection method based on dual-camera different-view imaging according to claim 7, characterized in that, The , The calibration logic is as follows: In step S3, the ROI of the template image under standard alignment is located using dual channels, and the original relative displacement at this point is calculated. Its opposite is the benchmark compensation amount.

9. A pattern and texture misalignment detection system based on dual-camera different-view imaging, characterized in that, The offset detection system is applied to the offset detection method according to any one of claims 1-8, wherein the pattern and texture offset detection system based on dual-camera different-view imaging includes: Camera video control module: used to automatically control the installed camera to perform intermittent video recording of the conveyed products, and send the information data collected by the camera to subsequent modules for calculation and processing; Displacement numerical calculation module: It is used to automatically receive information data collected by the camera control module, input the received information data into the model for displacement calculation, and then send the calculated data to the subsequent modules for judgment and processing; The regional value determination module is used to independently determine the calculated deviation error and automatically output the final detection conclusion and related data to support subsequent processing on the production line.

10. A pattern and texture misalignment detection system based on dual-camera heterogeneous view imaging according to claim 9, characterized in that, The camera shooting control module includes a camera shooting control unit, a template reference acquisition unit, an acquisition data preprocessing unit, and a data output unit. The camera shooting control unit is used to automatically control the intermittent operation and shutdown of the two cameras. The template reference acquisition unit is used to receive reference image information data acquired by the two cameras. The acquisition data preprocessing unit is used to preprocess the acquired data to remove invalid data such as blurry images.

Images