Ink-jet printing distance adjusting system and method based on computer vision

Abstract

The invention discloses an ink-jet printing distance adjusting system and method based on computer vision. The method comprises the following steps that image data of ink-jet points are collected in real time through a computer vision module; transmitting the acquired image data to a computer analysis module, and processing the image through an image processing algorithm; the computer analysis module carries out deviation calculation; the ink-jet spacing of each nozzle in the ink-jet printing equipment is adjusted in real time, and meanwhile, the priority distribution of ink-jet points is carried out; and inputting feedback data generated after each printing task is finished into a machine learning algorithm for training and optimization. Compared with the prior art, the method has the following advantages and effects that computer vision and machine learning technologies are combined, automatic and intelligent adjustment of the ink-jet printing equipment is achieved, the adjustment precision and the working efficiency of the ink-jet printing equipment are improved, and the adaptability of the ink-jet printing equipment in variable working environments is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of printing technology, and in particular to an inkjet printing spacing adjustment system and method based on computer vision. Background Art

[0002] With the continuous development of inkjet printing technology, especially in the fields of high precision and high quality requirements, such as image printing, publishing, advertising and electronic product production, the performance of inkjet printing equipment has gradually become an important guarantee for production efficiency and product quality. One of the core of inkjet printing equipment is the control of the precision of the nozzle and the inkjet spacing. The performance of the inkjet head directly affects the precise positioning of the inkjet point and the printing quality. However, the traditional inkjet spacing adjustment method has problems such as low precision and poor adaptability when dealing with complex images or changing working environments.

[0003] Currently, many inkjet printing devices adjust the inkjet head inkjet spacing manually or semi-automatically. These methods are usually adjusted based on experience or preset standards and rely on manual intervention or simplified automation processes. However, these methods are usually unable to effectively cope with changes in the inkjet point position under changing environmental conditions, resulting in large inkjet deviations, affecting printing effects, and thus reducing product quality and production efficiency.

[0004] Although computer vision technology has great potential in inkjet printing, in practical applications, the accuracy of image acquisition, the efficiency of image processing algorithms and the adaptive ability to adjust the inkjet spacing are still technical challenges that need to be solved urgently. Summary of the invention

[0005] The purpose of the present invention is to provide an inkjet printing spacing adjustment system and method based on computer vision to solve the problems raised in the background technology.

[0006] The above technical objectives of the present invention are achieved through the following technical solutions:

[0007] On the one hand, the present application provides a method for adjusting inkjet printing spacing based on computer vision, comprising the following steps:

[0008] S100, collecting image data of inkjet dots on the surface of the printed object in real time through a computer vision module installed on the inkjet printing device;

[0009] S200, transmitting the image data collected by the computer vision module to the computer analysis module, and the computer analysis module processes the image using an image processing algorithm;

[0010] S300, the computer analysis module performs deviation calculation based on the processed image to determine the deviation between the actual position of each inkjet point and the preset ideal position;

[0011] S400, adjusting the inkjet spacing of each nozzle in the inkjet printing device in real time according to the deviation calculated in step S300, and performing inkjet point priority allocation at the same time;

[0012] S500, during the printing process, monitor the position change of the inkjet point in real time to ensure that each adjustment can gradually reduce the deviation until the actual position of the inkjet point is highly consistent with the ideal position;

[0013] S600, the feedback data generated after each printing task is input into the machine learning algorithm for training and optimization, and then improve the adjustment efficiency based on the long-term accumulated feedback data to adapt to the complex and changing working environment.

[0014] By adopting the above technical solution, the present invention collects inkjet point image data on the surface of the printed object in real time on the inkjet printing equipment by installing a computer vision module, and can dynamically capture the actual distribution of the inkjet points, and reflect the relative position between the nozzle and the printed object surface in real time; because computer vision can obtain image data with high precision, the deviation calculation of the inkjet points is more accurate, and this real-time feedback mechanism can provide a direct basis for subsequent nozzle position adjustment, ensure the precise control of each nozzle during the printing process, and improve the overall quality of printing.

[0015] In the computer analysis module, the image data is further processed, and the deviation between the actual position of each inkjet point and the preset ideal position is calculated to effectively correct the offset of the inkjet point. The result of the deviation calculation provides specific parameters for the adjustment of the nozzle, which can not only improve the accuracy of the inkjet position, but also make fine adjustments in the inkjet spacing and inkjet flow of the nozzle to achieve a more stable printing effect; during the adjustment of the inkjet spacing of the nozzle, the system can correct the nozzle position in real time to ensure uniform inkjet quality. Through this coordinated adjustment, the inkjet equipment can maintain a high printing quality, especially in complex environments, and can effectively avoid common inkjet deviations, such as ink leakage or heavy ink problems.

[0016] In addition, the present invention can realize automatic learning of long-term accumulated experience by introducing a machine learning algorithm to train and optimize historical feedback data, and further improve the adjustment accuracy of the inkjet device under different environmental conditions.

[0017] It is further configured that the image processing process in step S200 includes image preprocessing, edge detection, image registration and image segmentation, specifically:

[0018] S201, preprocessing the image by using Gaussian filtering and median filtering, wherein Gaussian filtering smoothes the image by weighted averaging, and median filtering removes salt and pepper noise, and Gaussian filtering and median filtering are combined to improve image quality and ensure the effect of subsequent edge detection and image segmentation;

[0019] S202, performing edge detection, using the Sobel operator to take the image as input, obtaining the edge strength and direction of each pixel by calculating the horizontal and vertical gradients, and obtaining the preliminary position of the edge in the image;

[0020] After the Sobel operator is used to obtain the initial position of the edge in the image, the Canny algorithm is used to refine these edges through multi-level processing, and finally a detailed and accurate inkjet point edge is obtained, so that the edge detection result is more in line with the requirements of inkjet point positioning;

[0021] S203, performing image registration, aligning images collected from different viewing angles or at different times, so that key feature points in the images can be accurately matched to reduce geometric errors between images; the image registration process includes scale-invariant feature transformation, accelerated robust feature transformation and geometric transformation;

[0022] S204, after completing edge detection and image registration, the computer analysis module performs image segmentation to extract each inkjet point from the image; the computer analysis module preliminarily separates the inkjet point area and the background area in the image, and selects a suitable threshold T according to the grayscale histogram of the image. Each pixel point I (x, y) in the image will be compared with the threshold T:

[0023]

[0024] Among them, I seg (x, y)I represents the binary image after segmentation, a value of 1 represents the inkjet dot area, and a value of 0 represents the background area.

[0025] By adopting the above technical solution, the Gaussian filter and the median filter are combined to effectively remove noise and smooth the image. The Gaussian filter smoothes the image by weighted averaging, which is suitable for processing low-frequency noise in the image; while the median filter effectively removes the salt and pepper noise, making the image clearer; this processing method ensures that the subsequent edge detection and image segmentation algorithms can be performed on the basis of a clearer image, thereby improving the accuracy of subsequent steps and reducing the probability of false detection. The Sobel operator is combined with the Canny edge detection algorithm to make the edges of the inkjet points in the image more accurate. The Canny algorithm effectively improves the accuracy of edge detection through multiple steps such as Gaussian filtering, gradient calculation, non-maximum suppression and double threshold processing. Detailed edge extraction is the key to the subsequent inkjet point positioning, which ensures that the position of each inkjet point can be accurately identified and provides reliable data support for deviation calculation. In the image registration stage, the use of scale-invariant feature transformation and accelerated robust feature transformation algorithms can achieve high-precision image alignment and eliminate geometric errors between images collected from different perspectives and at different times. This step is crucial to ensure accurate matching of key feature points in each image, especially in the case of complex or irregular inkjet point distribution. Image registration can effectively improve the accuracy of subsequent image analysis and avoid the accumulation of deviations caused by registration errors. According to the grayscale histogram of the image, the appropriate threshold T is selected for image segmentation to ensure effective separation of the inkjet point area and the background area; image segmentation can extract the inkjet points in the target area, avoid interference from background noise, accurately locate the position of each inkjet point, and provide a clean data source for deviation calculation, thereby improving the accuracy of deviation calculation.

[0026] A further configuration is: in the step S201, after the preprocessing is completed, the computer analysis module equalizes the image brightness through histogram equalization to enhance the image contrast, making the details more obvious, especially in complex or low-contrast images, and can highlight the edge information of the inkjet dots for use in step S202.

[0027] By adopting the above technical solution,

[0028] By equalizing the brightness of the image through histogram equalization, the contrast of the image can be effectively improved, especially in complex or low-contrast images, and the details of the image can be enhanced; this processing step is particularly important for highlighting the edge information of the inkjet point, ensuring that the edge detection and inkjet point positioning in the subsequent steps are more accurate.

[0029] Further configuration is: in the step S300, after each inkjet point is accurately extracted from the image by the image processing algorithm, the computer analysis module performs deviation calculation to determine the deviation between the actual position of each inkjet point and the preset ideal position, and the deviation calculation includes Euclidean distance calculation, specifically:

[0030] Based on the actual position of each inkjet point and the preset ideal position, the Euclidean distance is calculated as a metric of deviation;

[0031]

[0032] Among them, D euclid Indicates the deviation of the inkjet point, x actual and actual is the actual position coordinate of the inkjet point, x ideal and ideal The ideal position preset for the inkjet dots.

[0033] By adopting the above technical solution and using Euclidean distance as the deviation metric, the deviation between the actual position and the ideal position of each inkjet point can be effectively measured. The Euclidean distance can accurately reflect the position difference of each inkjet point and provide precise parameters for the adjustment of the nozzle. This method can effectively calculate the position deviation of each inkjet point, thereby guiding the inkjet device to make necessary adjustments in the physical space.

[0034] Further configuration is: in the step S300, after the computer analysis module performs the Euclidean distance calculation, the deviation is further refined by weighted deviation calculation, and the weighted deviation calculation is performed by assigning a weight coefficient w to the deviation of each inkjet point, thereby correcting the deviation according to the deviation size and importance of the inkjet point, and the weight coefficient w is determined based on the deviation size, the importance of the inkjet point, the inkjet head state and environmental factors; the weighted deviation calculation formula is:

[0035] D weighted =w×D euclid

[0036] Among them, D weighted is the weighted deviation, D euclid is the deviation obtained from the Euclidean distance calculation.

[0037] By adopting the above technical solution, the strategy of deviation correction is further optimized through weighted deviation calculation, combined with the deviation size, the importance of the inkjet point and the working status of the inkjet head. The weighted deviation calculation can assign priorities to the deviations between the inkjet points and the ideal positions according to the spatial importance of the inkjet points, thereby more effectively adjusting the positions of key inkjet points.

[0038] It is further configured that, in the step S400, the adjustment of the inkjet spacing of the nozzle includes the adjustment of the physical position of the nozzle, the adjustment of the inkjet flow rate and voltage, and the fine adjustment of the inkjet angle.

[0039] By adopting the above technical solution, through adjusting the physical position of the nozzle, adjusting the inkjet flow and voltage, and fine-tuning the parameters of the inkjet angle, the working state of the nozzle can be accurately adjusted, and the accuracy of the inkjet process can be ensured, reducing inkjet deviation and other common problems, and significantly improving printing quality.

[0040] It is further configured that, in the step S400, the inkjet point priority allocation is specifically as follows:

[0041] S401, the inkjet points in the center area of ​​the image have a great impact on the quality of the entire image, and the inkjet points in the center area need to be adjusted with high priority, especially when the deviation is large, correction must be given priority;

[0042] S402, the inkjet points located in the edge area of ​​the image have little effect on the quality of the overall image, and the inkjet points in the edge area of ​​the image can be corrected as necessary after the adjustment of the central area of ​​the image is completed;

[0043] S403: In order to quantify the priority of the inkjet point, a weight w may be assigned based on the spatial position of the inkjet point, the deviation size, and the state of the inkjet head. priority ; Among them, the inkjet points located in the center area of ​​the image will receive a higher weight, and the inkjet points located in the edge area of ​​the image will receive a lower weight:

[0044]

[0045] Among them, w priority is the priority weight of the inkjet point, α is an adjustment coefficient used to balance the weight difference of the inkjet points in the center and edge areas of the image, and D distance is the distance from the inkjet point to the center of the image.

[0046] By adopting the above technical scheme, priority weights can be allocated according to the spatial importance of the image position of the inkjet point; the inkjet points located in the central area usually have a greater impact on the image quality, and therefore need to be adjusted first. Through reasonable priority allocation, it can be ensured that the inkjet point deviation in the central area of ​​the image is corrected in time, and the inkjet points in the edge area can be corrected after the adjustment of the central area is completed; this method can effectively improve the working efficiency of the inkjet equipment.

[0047] A further configuration is that the machine learning algorithm in step S600 adopts a random forest algorithm to satisfy the nonlinear relationship between the deviation of the inkjet point and the adjustment of the inkjet spacing of the nozzle.

[0048] By adopting the above technical solution and introducing the random forest algorithm, the present invention can handle the nonlinear relationship between the deviation of the inkjet point and the inkjet spacing of the nozzle; the random forest algorithm can learn the adjustment rules of the inkjet equipment by training a large amount of historical feedback data, and dynamically optimize the adjustment strategy of the inkjet spacing according to different environments and working conditions, thereby adaptively optimizing the operating parameters of the inkjet equipment, reducing manual intervention, and automatically adjusting the inkjet equipment so that it can maintain high accuracy and stability in different environments.

[0049] On the other hand, the present application also provides an inkjet printing spacing adjustment system based on computer vision, including a computer vision module installed on the inkjet printing equipment and a computer analysis module that communicates data with the computer vision module.

[0050] The present invention has the following beneficial effects:

[0051] The present invention combines computer vision and machine learning technologies to achieve automated and intelligent adjustment of inkjet printing equipment, which not only improves the adjustment accuracy and working efficiency of the inkjet equipment, but also enhances its adaptability in changing working environments; through efficient image processing algorithms, accurate deviation calculation and adaptive optimization mechanisms, the present invention provides a high-precision, high-efficiency and low-maintenance solution for the inkjet printing industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow chart;

[0053] Figure 2 This is a comparison table of the deviation after applying the traditional method and the deviation after applying the present invention in Example 1;

[0054] Figure 3 A table schematic diagram of the inkjet point priority allocation effect in Example 1;

[0055] Figure 4 A table schematically showing the effect of inkjet point priority allocation in Example 2. DETAILED DESCRIPTION

[0056] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0057] As attached Figure 1 As shown;

[0058] This embodiment discloses a method for adjusting inkjet printing spacing based on computer vision, comprising the following steps:

[0059] S100, the image data of the inkjet dots on the surface of the printed object is collected in real time by a computer vision module installed on the inkjet printing device; the computer vision module uses multiple high-resolution CCD cameras or CMOS cameras and a light source unit; in the inkjet printing device, the computer vision module collects the image data of the inkjet dots on the surface of the printed object in real time by setting multiple high-resolution CCD cameras or CMOS cameras. Each camera is precisely arranged at different angles to ensure that the inkjet dots at different positions can be clearly captured. The light source unit provides uniform illumination for each camera, eliminating image quality problems caused by uneven illumination or environmental factors, thereby obtaining high-quality and clear inkjet dot images. The data after image acquisition will be transmitted to the computer analysis module in real time through a high-bandwidth data transmission channel (such as an optical fiber connection or a dedicated network protocol). During the data transmission process, a compression algorithm is used to ensure that the image data reduces bandwidth consumption while ensuring the transmission speed, and ensures the integrity and accuracy of the data. The computer analysis module receives the image and prepares for the image processing in step S200.

[0060] S200, transmitting the image data collected by the computer vision module to the computer analysis module, and the computer analysis module processes the image using an image processing algorithm;

[0061] S300, the computer analysis module performs deviation calculation based on the processed image to determine the deviation between the actual position of each inkjet point and the preset ideal position;

[0062] S400, adjusting the inkjet spacing of each nozzle in the inkjet printing device in real time according to the deviation calculated in step S300, and performing inkjet point priority allocation at the same time;

[0063] S500, during the printing process, monitor the position change of the inkjet point in real time to ensure that each adjustment can gradually reduce the deviation until the actual position of the inkjet point is highly consistent with the ideal position;

[0064] S600, the feedback data generated after each printing task is input into the machine learning algorithm for training and optimization, and then improve the adjustment efficiency based on the long-term accumulated feedback data to adapt to the complex and changing working environment.

[0065] The image processing process in step S200 includes image preprocessing, edge detection, image registration and image segmentation, specifically:

[0066] S201, preprocessing the image by using Gaussian filtering and median filtering, wherein Gaussian filtering smoothes the image by weighted averaging, and median filtering removes salt and pepper noise, and Gaussian filtering and median filtering are combined to improve image quality and ensure the effect of subsequent edge detection and image segmentation;

[0067] S202, performing edge detection, using the Sobel operator to take the image as input, obtaining the edge strength and direction of each pixel by calculating the horizontal and vertical gradients, and obtaining the preliminary position of the edge in the image; however, the Sobel operator is prone to generate noise, especially in the case of poor image quality or complex background, the definition of the edge may not be clear enough;

[0068] After the Sobel operator is used to obtain the initial position of the edge in the image, the Canny algorithm is used to refine these edges through multi-level processing (including Gaussian filtering, gradient calculation, non-maximum suppression and double threshold processing), avoiding the noise and pseudo-edge problems that are prone to occur in the Sobel operator, and finally obtaining detailed and accurate inkjet point edges, so that the edge detection results are more in line with the requirements of inkjet point positioning;

[0069] S203. Since the camera may have an angle error during the inkjet printing process, or the captured image may have geometric deformation due to the uneven surface of the printed object, image registration must be performed to align images collected from different viewing angles or at different times so that key feature points in the images can be accurately matched to reduce the geometric error between images; the image registration process includes scale-invariant feature transformation, accelerated robust feature transformation and geometric transformation;

[0070] Geometric transformation can rotate, translate or perspective transform the image to eliminate angle errors or viewing angle differences. For example, an image can be projected onto a standard plane through affine transformation or perspective transformation to ensure accurate alignment of inkjet dots.

[0071] S204, after completing edge detection and image registration, the computer analysis module performs image segmentation to extract each inkjet point from the image; the computer analysis module preliminarily separates the inkjet point area and the background area in the image, and selects a suitable threshold T according to the grayscale histogram of the image. Each pixel point I (x, y) in the image will be compared with the threshold T:

[0072]

[0073] Among them, I seg (x, y)I represents the binary image after segmentation, a value of 1 represents the inkjet dot area, and a value of 0 represents the background area.

[0074] Among them, in step S201, after the preprocessing is completed, the computer analysis module equalizes the image brightness through histogram equalization, improves the image contrast, makes the details more obvious, especially in complex or low-contrast images, can highlight the edge information of the inkjet point for use in step S202.

[0075] Among them, in step S300, after each inkjet point is accurately extracted from the image by the image processing algorithm, the computer analysis module performs deviation calculation to determine the deviation between the actual position of each inkjet point and the preset ideal position. The deviation calculation includes Euclidean distance calculation, specifically:

[0076] Based on the actual position of each inkjet point and the preset ideal position, the Euclidean distance is calculated as a metric of deviation;

[0077]

[0078] Among them, D euclid Indicates the deviation of the inkjet point, x actual and actual is the actual position coordinate of the inkjet point, x ideal and ideal The ideal position preset for the inkjet dots.

[0079] Among them, in step S300, after the computer analysis module performs the Euclidean distance calculation, the deviation is further refined by weighted deviation calculation. The weighted deviation calculation is performed by assigning a weight coefficient w to the deviation of each inkjet point, thereby correcting the deviation according to the deviation size and importance of the inkjet point. The weight coefficient w is determined based on the deviation size, the importance of the inkjet point, the inkjet head state and environmental factors; the weighted deviation calculation formula is:

[0080] D weighted =w×D euclid

[0081] Among them, D weighted is the weighted deviation, D euclid is the deviation obtained from the Euclidean distance calculation.

[0082] After completing step S300, the computer analysis module has obtained the deviation information of each inkjet point through Euclidean distance calculation and weighted deviation calculation. At this point, the Euclidean distance calculation can clearly understand the gap between each inkjet point and its ideal position; in order to achieve the best inkjet spacing, the inkjet printing equipment needs to be adjusted according to this deviation information.

[0083] In step S400, the adjustment of the inkjet spacing of the nozzle includes the adjustment of the physical position of the nozzle, the adjustment of the inkjet flow rate and voltage, and the fine adjustment of the inkjet angle.

[0084] In step S400, the inkjet point priority allocation is specifically as follows:

[0085] S401. The inkjet dots in the center of the image have a great impact on the quality of the overall image. The details, contrast and color of the center usually determine the visual effect of the image. Therefore, ensuring the precise alignment of the inkjet dots in these areas is crucial for the entire printing process. The inkjet dots in the center area need to be adjusted with a higher priority, especially when the deviation is large, correction must be given priority;

[0086] S402, the inkjet dots in the edge area of ​​the image have little effect on the quality of the overall image, and a slight deviation of the inkjet dots may not significantly affect the quality of the overall image, especially in the edge area, where the details of color and graphics usually do not play a decisive role in the overall printing quality as in the center area. Therefore, the inkjet dots in the edge area of ​​the image can be corrected as necessary after the adjustment of the center area of ​​the image is completed;

[0087] S403: In order to quantify the priority of the inkjet point, a weight w may be assigned based on the spatial position of the inkjet point, the deviation size, and the state of the inkjet head. priority ; Among them, the inkjet points located in the center area of ​​the image will receive a higher weight, and the inkjet points located in the edge area of ​​the image will receive a lower weight:

[0088]

[0089] Among them, w priority is the priority weight of the inkjet point, α is an adjustment coefficient used to balance the weight difference of the inkjet points in the center and edge areas of the image, and D distance is the distance from the inkjet point to the center of the image;

[0090] Then, the inkjet points in the center area will get a higher priority correction, while the inkjet points at the edge will get a lower priority correction. Thus, it can ensure that the priority of the inkjet points is reasonably allocated, and the inkjet spacing of the nozzles can be dynamically adjusted according to the size of the deviation.

[0091] The machine learning algorithm in step S600 adopts a random forest algorithm to satisfy the nonlinear relationship between the deviation of the inkjet point and the adjustment of the inkjet spacing of the nozzle.

[0092] In addition, as the data dimension increases (for example, including multiple variables such as the working status of the inkjet head and environmental factors), random forests can effectively process high-dimensional data and avoid the dimensionality curse problem; random forests can effectively learn when there is a lot of noise and outliers in the computer vision module and computer analysis module, and have good adaptability to data in real production environments. Since the adjustment of inkjet printing equipment involves multiple factors, random forests can learn the complex relationship between various factors through the integration of multiple decision trees, so as to make more accurate inkjet adjustments.

[0093] This embodiment also discloses an inkjet printing spacing adjustment system based on computer vision, including a computer vision module installed on the inkjet printing device and a computer analysis module that communicates data with the computer vision module.

[0094] Example 1

[0095] When a printing company was printing high-precision artworks, the deformation of the printed objects and the fluctuation of ambient temperature and humidity caused the inkjet point position deviation in the center area of ​​the printed pattern to reach ±50μm, and the deviation in the edge area to reach ±80μm, resulting in a scrap rate of up to 12%. After applying the inkjet printing spacing adjustment method based on computer vision proposed in the present invention, the printing accuracy was significantly improved through real-time visual feedback and machine learning optimization.

[0096] The hardware of the computer vision module in step S100 adopts a Basler ace 2 camera, which has 5 million pixels and a frame rate of 120fps, and also adopts a ring LED light source with a color temperature of 5500K.

[0097] In step S201, the Gaussian filter is configured with σ=1.2 and a 3×3 convolution kernel, and the median filter uses a 5×5 window.

[0098] In step S202, the Canny algorithm is configured with a low threshold of 50, a high threshold of 150, and three edge tracking iterations.

[0099] In step S204, the image segmentation threshold T is set to 120, which is determined based on the grayscale histogram statistics of 1000 samples.

[0100] The deviation after applying the traditional method in step S300 and the deviation after applying the present invention are shown in the attached figure. Figure 2 As shown in the table;

[0101] After the inkjet spacing of the nozzle is adjusted in step S400, the inkjet point priority allocation effect is as shown in the attached figure. Figure 3 As shown in the table;

[0102] In step S600, 10,000 sets of printing task data are used;

[0103] After three months of production verification, the scrap rate dropped from 12% to 1.8%, the average positioning accuracy increased to ±6.3μm, and the changeover time was shortened by 42%.

[0104] Example 2

[0105] When a high-end packaging printing factory produces luxury gift boxes, it needs to print complex gradient patterns on matte paper. However, the inkjet head causes the inkjet point position to shift due to mechanical vibration, resulting in color misalignment. Specifically, the overlap deviation of cyan and magenta is as high as 0.3mm. After applying the inkjet printing spacing adjustment method based on computer vision proposed in the present invention, the printing waste rate is reduced from 15% to 0.5%.

[0106] The hardware of the computer vision module in step S100 uses an industrial camera with 20 million pixels and a frame rate of 120fps, and is equipped with a ring-shaped LED cold light source with a color temperature of 5500K;

[0107] In step S201, the Gaussian filter is configured with σ=1.5 and 5×5 Gaussian kernel to eliminate the texture noise of the printed matter, so that the noise standard deviation is reduced from 25.3 to 8.7; the median filter uses a 7×7 window to eliminate most of the salt and pepper noise caused by nozzle blockage, and the number of noise points is reduced from 112 / cm 2 Reduced to 3 / cm 2 ;

[0108] In step S202, the Sobel operator preliminarily locates the edges of 2500 inkjet dots; the Canny algorithm is configured with a low threshold of 50 and a high threshold of 150;

[0109] In step S204, the image segmentation threshold T is set to 180, and the segmentation accuracy reaches 99.2%;

[0110] In step S400, the inkjet dots in the center area are adjusted within 3ms, and the edge areas are delayed to 10ms for post-processing; the inkjet dot priority allocation effect is shown in the attached figure. Figure 4 As shown in the table;

[0111] In step S600, 50,000 sets of printing task data are used;

[0112] After three months of production verification, the printing waste rate dropped from 15% to 0.5%.

[0113] This specific embodiment is merely an explanation of the present invention and is not a limitation of the present invention. After reading this specification, those skilled in the art may make non-creative modifications to the present embodiment as needed. However, as long as they are within the scope of the claims of the present invention, they are protected by the patent law.

Claims

1. A method for adjusting inkjet printing spacing based on computer vision, characterized in that: The following steps are included: S100, collecting image data of inkjet dots on the surface of the printed object in real time through a computer vision module installed on the inkjet printing device; S200, transmitting the image data collected by the computer vision module to the computer analysis module, and the computer analysis module processes the image using an image processing algorithm; S300, the computer analysis module performs deviation calculation based on the processed image to determine the deviation between the actual position of each inkjet point and the preset ideal position; S400, adjusting the inkjet spacing of each nozzle in the inkjet printing device in real time according to the deviation calculated in step S300, and performing inkjet point priority allocation at the same time; S500, during the printing process, monitor the position change of the inkjet point in real time to ensure that each adjustment can gradually reduce the deviation until the actual position of the inkjet point is highly consistent with the ideal position; S600, the feedback data generated after each printing task is input into the machine learning algorithm for training and optimization, and then improve the adjustment efficiency based on the long-term accumulated feedback data to adapt to the complex and changing working environment.

2. The inkjet printing spacing adjustment method based on computer vision according to claim 1, characterized in that: The image processing process in step S200 includes image preprocessing, edge detection, image registration and image segmentation, specifically: S201, preprocessing the image by using Gaussian filtering and median filtering, wherein Gaussian filtering smoothes the image by weighted averaging, and median filtering removes salt and pepper noise, and Gaussian filtering and median filtering are combined to improve image quality and ensure the effect of subsequent edge detection and image segmentation; S202, performing edge detection, using the Sobel operator to take the image as input, obtaining the edge strength and direction of each pixel by calculating the horizontal and vertical gradients, and obtaining the preliminary position of the edge in the image; After the Sobel operator is used to obtain the preliminary position of the edge in the image, the Canny algorithm is used to refine these edges through multi-level processing, and finally a detailed and accurate inkjet dot edge is obtained, so that the edge detection result is more in line with the requirements of inkjet dot positioning; S203, performing image registration, aligning images collected from different viewing angles or at different times, so that key feature points in the images can be accurately matched to reduce geometric errors between images; the image registration process includes scale-invariant feature transformation, accelerated robust feature transformation and geometric transformation; S204, after completing edge detection and image registration, the computer analysis module performs image segmentation to extract each inkjet point from the image; the computer analysis module preliminarily separates the inkjet point area and the background area in the image, and selects a suitable threshold T according to the grayscale histogram of the image. Each pixel point I (x, y) in the image will be compared with the threshold T: Among them, I seg (x, y)I represents the binary image after segmentation, a value of 1 represents the inkjet dot area, and a value of 0 represents the background area.

3. The inkjet printing spacing adjustment method based on computer vision according to claim 2, characterized in that: In the step S201, after the preprocessing is completed, the computer analysis module equalizes the image brightness through histogram equalization to enhance the image contrast and make the details more obvious, especially in complex or low-contrast images, and can highlight the edge information of the inkjet dots for use in step S202.

4. The inkjet printing spacing adjustment method based on computer vision according to claim 1, characterized in that: In the step S300, after each inkjet point is accurately extracted from the image by the image processing algorithm, the computer analysis module performs deviation calculation to determine the deviation between the actual position of each inkjet point and the preset ideal position. The deviation calculation includes Euclidean distance calculation, specifically: Based on the actual position of each inkjet point and the preset ideal position, the Euclidean distance is calculated as a metric of deviation; Among them, D euclid Indicates the deviation of the inkjet point, x actual and actual is the actual position coordinate of the inkjet point, x ideal and ideal The ideal position preset for the inkjet dots.

5. The inkjet printing spacing adjustment method based on computer vision according to claim 4, characterized in that: In the step S300, after the computer analysis module performs the Euclidean distance calculation, the deviation is further refined by weighted deviation calculation. The weighted deviation calculation corrects the deviation according to the deviation size and importance of the inkjet point by assigning a weight coefficient w to the deviation of each inkjet point. The weight coefficient w is determined based on the deviation size, the importance of the inkjet point, the inkjet head state and environmental factors. The weighted deviation calculation formula is: D weighted =w×D euclid Among them, D weighted is the weighted deviation, D euclid is the deviation obtained from the Euclidean distance calculation.

6. The inkjet printing spacing adjustment method based on computer vision according to claim 1, characterized in that: In the step S400, the adjustment of the inkjet spacing of the nozzle includes the adjustment of the physical position of the nozzle, the adjustment of the inkjet flow rate and voltage, and the fine adjustment of the inkjet angle.

7. The inkjet printing spacing adjustment method based on computer vision according to claim 1, characterized in that: In the step S400, the inkjet point priority allocation is specifically as follows: S401, the inkjet points in the center area of ​​the image have a great impact on the quality of the entire image, and the inkjet points in the center area need to be adjusted with high priority, especially when the deviation is large, correction must be given priority; S402, the inkjet points located in the edge area of ​​the image have little effect on the quality of the overall image, and the inkjet points in the edge area of ​​the image can be corrected as necessary after the adjustment of the central area of ​​the image is completed; S403: In order to quantify the priority of the inkjet point, a weight w may be assigned based on the spatial position of the inkjet point, the deviation size, and the state of the inkjet head. priority ; Among them, the inkjet points located in the center area of ​​the image will receive a higher weight, and the inkjet points located in the edge area of ​​the image will receive a lower weight: Among them, w priority is the priority weight of the inkjet point, α is an adjustment coefficient used to balance the weight difference of the inkjet points in the center and edge areas of the image, and D distance is the distance from the inkjet point to the center of the image.

8. The inkjet printing spacing adjustment method based on computer vision according to claim 1, characterized in that: The machine learning algorithm in step S600 adopts a random forest algorithm to satisfy the nonlinear relationship between the deviation of the inkjet point and the adjustment of the inkjet spacing of the nozzle.

9. An inkjet printing spacing adjustment system based on computer vision, applied to the inkjet printing spacing adjustment method based on computer vision according to any one of claims 1 to 8, characterized in that: The invention comprises a computer vision module installed on the inkjet printing device and a computer analysis module which is interconnected with the computer vision module in data communication.

Images