Ink droplet image observation operation system based on ink-jet printer

Abstract

The invention discloses an ink droplet image observation operation system based on an ink-jet printer. The system comprises an ink droplet video acquisition module; a video preprocessing module; a deep learning convolutional neural network module; an artificial intelligence automatic parameter optimization module; a deep learning convolutional neural network and an artificial intelligence automatic parameter optimization technology are adopted, an ink droplet detection precision algorithm and an ink droplet detection speed optimization algorithm are put forward on the basis of the deep learning convolutional neural network, convolution is performed on an ink droplet image through the deep learning convolutional neural network to obtain ink droplet features, and then, the ink droplet features are obtained. According to the method, the accuracy and the speed of ink droplets are obtained through an ink droplet detection accuracy algorithm and an ink droplet detection speed optimization algorithm for ink droplet characteristics, an automatic parameter optimization scheme is constructed through an artificial intelligence automatic parameter optimization technology, and the quality of the ink droplets is scored, so that the quality numerical value of the ink droplets is obtained, and the quality of the ink droplets is controlled.

Description

Technical Field

[0001] The invention relates to the technical field of inkjet printers, and in particular to an operating system based on inkjet printer ink droplet image observation. Background Art

[0002] At present, driven by the increasing demand for information printing on product packaging in the food and beverage, pharmaceutical, electronics, automotive and other industries, the inkjet printer market is experiencing a sustained growth phase. The domestic market demand for inkjet printers will become greater and greater. The research and development of high-tech inkjet printers is of great significance, which is conducive to the in-depth promotion of technologies such as artificial intelligence, intelligent equipment, multi-sensors, and the full life cycle of industrial products. It has important practical significance to improve the automation level of the full-automatic packaging production line, which will greatly promote the traceability management capabilities of my country's industrial products throughout the life cycle, and promote the rapid development of my country's traceability code high-speed inkjet printer technology.

[0003] However, the high-speed ink flow breaks into ink droplets under the disturbance provided by the piezoelectric ceramic. When the breaking condition is not ideal, it will produce satellite ink droplets or cause the charge to be transferred between ink droplets, thus causing the quality of the ink droplets to deteriorate. The breaking is affected by many factors such as nozzle aperture, ink flow speed, transducer frequency, transducer amplitude, ink viscosity, etc., which will cause problems such as poor inkjet printing effect and low quality. This requires users to observe the quality of the printing next to the inkjet printer from time to time, and then conduct on-site debugging, which increases time and labor costs. Summary of the invention

[0004] The purpose of the present invention is to provide an operating system based on inkjet printer ink droplet image observation to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: an operating system based on inkjet printer ink droplet image observation, characterized in that it includes:

[0006] The ink drop video acquisition module, based on the OpenCv function library, captures the video stream from the camera and displays the video frames on the computer;

[0007] The video preprocessing module, after reading the video, converts the color image into a grayscale image based on OpenCv, and binarizes the grayscale image to obtain a binary image;

[0008] Deep learning convolutional neural network module, which detects ink droplets in the video based on the ink droplet detection accuracy algorithm and ink droplet detection speed optimization algorithm based on deep learning;

[0009] The artificial intelligence automatic parameter optimization module builds an automatic parameter optimization solution based on the ink droplet automatic acquisition algorithm and the ink droplet automatic evaluation algorithm to compare and score the current ink droplet characteristics.

[0010] Furthermore, the ink drop video acquisition module includes a computer and a camera. Click the camera opening button to connect the computer to the camera. Click the video recording button to record the ink drop. Click the stop recording button to automatically stop the computer from recording and save it to the path specified by the user.

[0011] Furthermore, the video preprocessing module processes the video image of the ink droplets by clicking the real-time ink droplet processing button, and then clicks the ink droplet processing result button, and the computer interface displays the characteristic analysis results of the ink droplets in the video frame several seconds ago.

[0012] Furthermore, the ink drop video acquisition module also includes an ink drop video playback button and a screenshot button, the ink drop video playback button is used to reload and play the recorded ink drop video;

[0013] The screenshot button is used to take screenshots while the computer is playing the recorded ink drop video and save them to the path specified by the user.

[0014] Furthermore, the ink drop detection accuracy algorithm includes: a post-processing optimization algorithm BOPP based on edge lines, BOPP sets the positioning accuracy of the four edge lines of the ink drop detection positioning frame based on HOG features and random forests.

[0015] Furthermore, the ink drop detection speed optimization algorithm includes: combining the mobile image processing network MobileNet with the ink drop detection algorithm YOLO to construct a lightweight ink drop detection algorithm Mobile-YOLO with an automatic parameter optimization solution;

[0016] The backbone network structure of YOLO was replaced with depthwiseseparable convolution, and the network was compressed and redesigned.

[0017] Compared with the prior art, the beneficial effects of the present invention are: deep learning convolutional neural network and artificial intelligence automatic parameter optimization technology are adopted, and on the basis of the deep learning convolutional neural network, an ink droplet detection accuracy algorithm and an ink droplet detection speed optimization algorithm are proposed. The ink droplet features are obtained by convolving the ink droplet image through the deep learning convolutional neural network, and then the ink droplet features are obtained in the form of an ink droplet detection accuracy algorithm and an ink droplet detection speed optimization algorithm to obtain the accuracy and speed of the ink droplets. Through the artificial intelligence automatic parameter optimization technology, an automatic parameter optimization scheme is constructed to score the ink droplet quality, thereby obtaining the quality value of the ink droplets and realizing the control of the ink droplet quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a functional block diagram of the present invention;

[0019] Figure 2is a control flow chart of the present invention;

[0020] Figure 3 It is a functional control block diagram of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Example:

[0023] See also Figure 1-3 The present invention provides a technical solution: an operating system based on inkjet printer ink droplet image observation.

[0024] System function description diagram Figure 1 As shown, the steps include:

[0025] 1) The user clicks the "Open Camera" button on the interface and uses a high-frequency camera with LED imaging to quickly capture ink drop images and upload them to the computer;

[0026] The computer will automatically connect and start the camera. When the screen appears in the upper left corner of the UI interface, click the "Video Recording" button, the computer will start video recording, and the program will load and display the screen. After the recording is completed, click the Stop Recording button, the computer will automatically stop recording and save it to the path specified by the user.

[0027] 2) This UI interface uses two major technologies: deep learning convolutional neural network and artificial intelligence automatic parameter optimization. It will perform real-time analysis and processing on the ink droplets in the ink droplet video. Click the "Real-time processing of ink droplets" button, and the computer will process the video image of the ink droplets in real time. This process may take a few seconds. Then click the "Ink droplet processing results" button, and the interface will display the feature analysis results of the ink droplets in the video frame a few seconds ago, including but not limited to parameters such as mass, speed, volume, shape, and coverage.

[0028] 3) Click the "Ink Drop Video Playback" button, the computer will reload and play the recorded Ink Drop video for the user to view. Click the "Screenshot" button. While the computer is playing the recorded Ink Drop video, the user can press the "S" key to take a screenshot at any time. The screenshot will be automatically saved to the path specified by the user, and the prompt message "SCREENSHOT SAVED TO the path specified by the user" will be displayed.

[0029] 4) If the user needs to re-record the ink drop video, just repeat the above steps 1) and 2). This embodiment can be based on the system function description diagram and system function flow chart provided above. The present invention is mainly composed of a software system.

[0030] Software System

[0031] This embodiment can develop an inkjet printer UI operation interface based on OpenCV and Python according to the system functions provided above.

[0032] The software system is programmed in a modular way. It is mainly divided into four modules: ink drop video acquisition preprocessing algorithm, deep learning convolutional neural network and artificial intelligence automatic parameter optimization. Among them, deep learning convolutional neural network is the core module of this program and the key part of realizing ink drop image feature extraction. The flowchart of software implementation is as follows Figure 2 shown.

[0033] (1) Ink drop video acquisition

[0034] Video acquisition is an important step in obtaining ink droplet features. It mainly uses the OpenCv function library to capture video streams from the camera and display these video frames in the GUI. When the camera is turned off, the video capture is stopped and the camera resources are released.

[0035] (2) Preprocessing module

[0036] Data preprocessing mainly includes video reading and grayscale image binarization. First, image binarization is conducive to further image processing, making the image simpler, reducing the amount of data, and highlighting the outline of the target of interest. Secondly, to process and analyze binary images, the grayscale image must first be binarized to obtain a binary image.

[0037] (3) Deep Learning Convolutional Neural Network

[0038] In this module, we designed and proposed an ink drop detection accuracy algorithm and an ink drop detection speed optimization algorithm based on deep learning. The ink drop detection accuracy algorithm is achieved by deepening or widening the network layer, while the ink drop detection speed optimization algorithm was mentioned in the depthwise separable technology in MobileNet proposed by Google. It has greatly reduced the amount of neural network calculations, but it still cannot meet the real-time requirements of low-end devices, so it is combined with YOLO to optimize the algorithm. The principles of the above algorithms are as follows:

[0039] Ink droplet detection accuracy algorithm: A border-oriented post-processing optimization algorithm BOPP (Border-Oriented Post-Processing) is proposed to optimize the positioning accuracy of the four borders of the ink droplet detection positioning frame respectively, so as to achieve the purpose of optimizing the overall positioning frame. BOPP uses HOG features and random forests as the basis of the algorithm. Its features and regressors are highly interpretable and relatively fast. At the same time, it can be used to optimize most existing mainstream ink droplet detection algorithms and has strong scalability.

[0040] Ink droplet detection speed optimization algorithm: By combining the mobile image processing network MobileNet and the ink droplet detection algorithm YOLO, a lightweight ink droplet detection algorithm Mobile-YOLO is proposed to build an automatic parameter optimization solution based on deep learning. This project replaces the backbone network structure of YOLO with depthwiseseparable convolution, performs network compression processing, and redesigns a new network structure.

[0041] (4) Automatic parameter optimization using artificial intelligence

[0042] As the last module of the UI interface of the inkjet printer, it mainly combines the automatic ink droplet acquisition technology with the automatic ink droplet evaluation technology to build an automatic parameter optimization solution. It seeks the optimal parameter combination by continuously and automatically adjusting multiple parameters, and then compares and scores the current ink droplet characteristics (out of 10 points). The score can be used to express the current ink droplet quality in a concise and clear manner.

[0043] Integration and debugging of each module

[0044] The UI interface design of the inkjet printer is based on demand analysis. According to the actual needs in reality, it is designed by comprehensively considering the performance of system availability, accuracy, real-time performance, function expansion and convenient operation. In the actual process, by obtaining the current ink drop image and analyzing the current ink drop characteristics, it is convenient for users to judge the internal situation of the inkjet printer.

[0045] The ink drop video of the inkjet printer is obtained through the camera, and then the ink drop image is analyzed using the deep learning convolutional neural network technology and the automatic parameter optimization technology of artificial intelligence to obtain the ink drop characteristics and ink drop quality. These functions can be realized through a UI interface. The main program interface is shown in the figure below. The user interface is divided into two parts. The left side is the ink drop image acquisition part, and the right side is the ink drop image analysis and ink drop video capture part. Figure 3 shown.

[0046] The basic flow of the main program is: first click the "Open Camera" button, then click the "Video Recording" button. In a timer's message response function, read the ink drop image taken by the camera at a fixed time. Click the "Stop Recording" button to turn off the camera. At the same time, stop capturing the video and release the camera resources. The video will be automatically saved to the path specified by the user. Click the "Real-time Processing of Ink Drops" button, the computer will read the ink drop video, and analyze the ink drop image through deep learning convolutional neural network technology and artificial intelligence automatic parameter optimization technology. Click the "Ink Drop Processing Result" button to display the ink drop feature and quality values.

[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An operating system based on inkjet printer ink drop image observation, characterized in that: include: The ink drop video acquisition module, based on the OpenCv function library, captures the video stream from the camera and displays the video frames on the computer; The video preprocessing module, after reading the video, converts the color image into a grayscale image based on OpenCv, and binarizes the grayscale image to obtain a binary image; Deep learning convolutional neural network module, which detects ink droplets in the video based on the ink droplet detection accuracy algorithm and ink droplet detection speed optimization algorithm based on deep learning; The artificial intelligence automatic parameter optimization module builds an automatic parameter optimization solution based on the ink droplet automatic acquisition algorithm and the ink droplet automatic evaluation algorithm to compare and score the current ink droplet characteristics.

2. The system for observing and operating inkjet printer ink droplet images according to claim 1, characterized in that: The ink drop video acquisition module includes a computer and a camera. Click the camera opening button to connect the computer to the camera. Click the video recording button and the camera will record the ink drop. Click the stop recording button and the computer will automatically stop recording and save it to the path specified by the user.

3. The system for observing and operating inkjet printer ink droplet images according to claim 1, characterized in that: The video preprocessing module processes the video image of the ink droplets by clicking the real-time ink droplet processing button, and then clicks the ink droplet processing result button, and the computer interface displays the characteristic analysis results of the ink droplets in the video frame several seconds ago.

4. The system for observing and operating inkjet printer ink droplet images according to claim 2 is characterized in that: The ink drop video acquisition module also includes an ink drop video playback button and a screenshot button. The ink drop video playback button is used to reload and play the recorded ink drop video; The screenshot button is used to take screenshots while the computer is playing the recorded ink drop video and save them to the path specified by the user.

5. The system for observing and operating inkjet printer ink droplet images according to claim 1 is characterized in that: The ink drop detection accuracy algorithm includes: a post-processing optimization algorithm BOPP based on edge lines. BOPP sets the positioning accuracy of the four edge lines of the ink drop detection positioning frame based on HOG features and random forests.

6. The system for observing and operating inkjet printer ink droplet images according to claim 1, characterized in that: The ink drop detection speed optimization algorithm includes: combining the mobile image processing network MobileNet with the ink drop detection algorithm YOLO to build a lightweight ink drop detection algorithm Mobile-YOLO with an automatic parameter optimization solution; Replace the backbone network structure of YOLO with depthwiseseparable convolution, perform network compression, and redesign the network structure.

7. The system for observing and operating inkjet printer ink droplet images according to claim 1 is characterized in that: The deep learning convolutional neural network module detects ink droplets including but not limited to mass, speed, volume, shape, and coverage.

Images