A system for online monitoring of optical machine exposure patterns and its working method

Abstract

The present invention discloses a system for online monitoring of optical machine exposure graphics and its working method, including the following steps: setting a threshold range of similarity between the acquired image and the slice image; starting monitoring from the first frame of the acquired image and the slice image, and using the acquired image and the slice image to compare to determine the correctness of the exposure; if the first frame is exposed normally, starting from the next frame, directly determine the similarity between the subsequent frame and the previous frame, and if the similarity is within the allowed threshold range, determine that the current exposure is normal; if the similarity between a certain frame and the previous frame exceeds the threshold, it means that the current frame exposure may have errors, and after the detection is stable, continue to use the similarity between frames to determine the correctness of the exposure. The present invention monitors the push-scanning exposure process of ceramic photocuring additive manufacturing throughout the process, and can promptly alarm when abnormalities are found, and is applied to the application scenario of large-size photocuring ceramic additive manufacturing. The frame rate of monitoring is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of photocuring additive manufacturing process, and in particular to a system for online monitoring of photomechanical exposure patterns and a working method thereof. Background Art

[0002] With the development of 3D printing technology in my country, printing technology has gradually matured, and the size of printed products has gradually developed towards large-scale. Limited by the projection size of the optical machine, a single exposure of the optical machine cannot produce large-sized ceramic parts. The alternative method is to use a mobile optical machine or a mobile manufacturing platform, and dynamically update the projection graphics of the optical machine to achieve the purpose of large-scale exposure.

[0003] During this process, current ceramic photocuring additive manufacturing equipment does not monitor the exposure of the optical system of the equipment, and the actual exposure pattern of the optical system is an open-loop control. As the manufacturing time increases, the optical system is prone to incorrect projection patterns due to overheating, data cache failure, etc. If this defect is not discovered in time, it will lead to the failure of the printing task. Summary of the invention

[0004] The object of the present invention is to provide a system for online monitoring of light machine exposure patterns and a working method thereof, so as to solve the above-mentioned problems existing in the prior art.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] A working method for online monitoring of light machine exposure patterns comprises the following steps:

[0007] S100, setting a similarity threshold range between the collected image and the sliced ​​image;

[0008] S200, starting monitoring from the first frame of the acquired image and the slice image, and judging the correctness of the exposure by comparing the acquired image with the slice image;

[0009] S300, if the first frame is exposed normally, then starting from the next frame, directly determine the similarity between the subsequent frame and the previous frame, and if the similarity is within the allowed threshold range, determine that the current exposure is normal;

[0010] S400, if the similarity between a certain frame and the previous frame exceeds a threshold, it indicates that the current frame may have an exposure error, and subsequent frames are compared with the slice image to determine whether the exposure is correct;

[0011] S500: After the detection is stable, continue to use the similarity between frames to determine the correctness of exposure.

[0012] In some specific embodiments, the method of comparing the acquired image with the slice image is specifically:

[0013] After acquiring the monitoring image of the optical system through the monitoring camera, the image is corrected by using image processing technology to correct the side view exposure image to the front view angle, and obtain the exposure pattern under the front view angle; according to the position of the optical machine relative to the manufacturing plane, the true value of the exposure pattern is segmented from the entire slice image to determine whether the current exposure pattern is wrong.

[0014] In some specific embodiments, the step of acquiring an image includes:

[0015] S210, installing the image acquisition device and the optical machine on the same bracket and moving them simultaneously;

[0016] S220, recording the moving distance of the bracket using a position sensor device;

[0017] S230, counting the number of pulses returned by the position sensor device through the control system, and sending a synchronous refresh signal to the monitoring camera and the optical machine after a certain number of pulses are counted;

[0018] S240. According to the push-scan mechanism of the optical machine, the image is cut in the push-scan direction in units of pixels, and the monitoring camera obtains the real-time exposure image of the optical machine.

[0019] In some specific embodiments, the acquired exposure image and the exposure pattern under the normal viewing angle are scaled to the minimum resolution, and a deep learning network is used to determine whether the current exposure pattern is correct.

[0020] In some specific embodiments, the similarity algorithm between frames in S500 is specifically:

[0021] S510, representing the current frame image and the previous frame image as Mi and Mi-1 respectively;

[0022] S520, scaling and interpolating the current frame image matrix and the previous frame image matrix to reduce the resolution;

[0023] S530, adjusting the current frame image matrix and the previous frame image matrix into a square matrix;

[0024] S540, calculating the difference and eigenvalues ​​of the current frame image matrix and the previous frame image matrix.

[0025] A system for online monitoring of light machine exposure patterns based on the same invention comprises:

[0026] An image acquisition module, used for obtaining acquisition images and slice images;

[0027] A position acquisition module, used to acquire the position of the optical machine relative to the manufacturing plane;

[0028] An image analysis module, used for analyzing the status of the acquired image and the sliced ​​image;

[0029] The policy execution module is used to execute the commands of the image analysis module.

[0030] The beneficial effects of the present invention are:

[0031] The present invention discloses a system for online monitoring of optical machine exposure graphics and its working method, including the following steps: setting a threshold range of similarity between an acquisition image and a slice image; starting monitoring from the first frame of the acquisition image and the slice image, and using the acquisition image to compare with the slice image to determine the correctness of the exposure; if the first frame is exposed normally, starting from the next frame, directly determine the similarity between the subsequent frame and the previous frame, and if the similarity is within the allowed threshold range, determine that the current exposure is normal; if the similarity between a certain frame and the previous frame exceeds the threshold, it means that the current frame exposure may be wrong, and then several frames continue to use the method of comparing with the slice image to determine whether the exposure is correct; after the detection is stable, continue to use the similarity between frames to determine the correctness of the exposure. The present invention monitors the push-scanning exposure process of ceramic photocuring additive manufacturing throughout the process, and can promptly alarm when abnormalities are found, and is applied to the application scenario of large-size photocuring ceramic additive manufacturing. The frame rate of monitoring is greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of a working method for online monitoring of light machine exposure patterns of the present invention;

[0033] Figure 2 It is a structural diagram of another embodiment of a working method for online monitoring of light machine exposure patterns of the present invention;

[0034] Figure 3 It is a structural schematic diagram of another embodiment of a working method for online monitoring of light machine exposure patterns of the present invention;

[0035] Figure 4 It is a structural block diagram of a system for online monitoring of light machine exposure patterns of the present invention;

[0036] Figure 5 It is a flow chart of another embodiment of a working method for online monitoring of light machine exposure patterns of the present invention. DETAILED DESCRIPTION

[0037] 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. Examples of the embodiments are shown in the drawings, where the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain the present invention, and cannot be understood as limiting the present invention. In the description of the present invention, it should be understood that the orientation or position relationship indicated by the terms "top", "bottom", "inside", "outside", "axis", "circumferential" and the like is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention or simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, “plurality” means two or more than two, unless otherwise clearly and specifically defined.

[0038] In the present invention, unless otherwise clearly specified and limited, the terms "install", "connect", "connect", "fix", "join", "hinge", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0039] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", "one specific embodiment" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0040] Reference Figure 1 , Figure 2 , Figure 3 , Figure 4 and Figure 5 A working method for online monitoring of light machine exposure patterns is shown, comprising the following steps:

[0041] S100: Setting a similarity threshold range between the acquired image and the sliced ​​image.

[0042] S200 , monitoring starts from the first frame of the acquired image and the slice image, and correctness of exposure is determined by comparing the acquired image with the slice image.

[0043] S300: If the exposure of the first frame is normal, then starting from the next frame, directly determine the similarity between the subsequent frame and the previous frame. If the similarity is within the allowed threshold range, determine that the current exposure is normal.

[0044] S400: If the similarity between a certain frame and the previous frame exceeds a threshold, it indicates that the current frame may have exposure errors, and subsequent frames continue to be compared with the slice image to determine whether the exposure is correct.

[0045] S500: After the detection is stable, continue to use the similarity between frames to determine the correctness of exposure.

[0046] It should be noted here that in the process of large-scale ceramic photocuring additive manufacturing, the optical machine needs to perform several push-scans to complete the exposure of the entire image. In each push-scan process, the position of the optical machine relative to the vertical push-scan direction is a fixed value, and the position in the push-scan direction changes dynamically. The system needs to segment the projection image from the entire slice image according to the relative position of the optical machine, and update the projection image in real time during the movement of the optical machine.

[0047] In the present invention, the monitoring camera is installed on the side of the optical system. In order to shield the interference of the ambient light, a 405nm narrow-band light-transmitting sheet is installed at the front end of the camera lens. The characteristic of this light-transmitting sheet is that only the light of 405nm wavelength emitted by the optical system of the ceramic additive manufacturing equipment can pass through the light-transmitting sheet and enter the monitoring camera after being focused by the industrial lens. Light of other wavelengths is filtered out, so that the camera only shoots the projection image of the optical system. The image in the optical machine push-scan process is a continuously changing process, that is, a certain similarity is maintained between frames. By comparing this similarity, it can be judged whether the image exposure is wrong. Based on this, the technical route given by the present invention is as follows: starting from the first frame, the correctness of the exposure is judged by comparing the collected image with the slice image. If the first frame is exposed normally, starting from the following frames, the similarity between the subsequent frame and the previous frame is directly judged. If the similarity is within the allowable threshold range, it is judged that the current exposure is normal. If the similarity between a certain frame and the previous frame exceeds the threshold, it means that the current frame exposure may be wrong. After that, several frames continue to use the method of comparing with the slice image to judge whether the exposure is correct. After the detection is stable, the similarity between frames is used to determine the correctness of the exposure.

[0048] In some specific embodiments, the method of comparing the acquired image with the slice image is specifically:

[0049] Reference Figure 2 As shown, after the monitoring image of the optical system is acquired by the monitoring camera, the image is corrected by using image processing technology, the side view exposure image is corrected to the front view angle, and the exposure pattern under the front view angle is acquired; according to the position of the optical machine relative to the manufacturing plane, the true value of the exposure pattern is segmented from the entire slice image to determine whether the current exposure pattern is wrong. It should be noted here that after the monitoring camera acquires the monitoring image of the optical system, the image is corrected by using image processing technology, and the side view exposure image is corrected to the front view angle. Image correction is to calculate the correction matrix between the side view image and the front view image through a pair of standard images. According to the properties of plane projective mapping, a plane projective transformation is a linear transformation of its homogeneous vectors in 3 dimensions. This linear transformation relationship can be expressed by a non-singular 3x3 matrix:

[0050]

[0051] This method uses optical projection to mark the printing area, and then uses image processing methods to calculate the coordinates of the four corner points of the rectangular printing area. These four coordinate values ​​are combined with the corner point coordinates of the true image to establish a set of equations to calculate the transformation matrix. The transformation matrix is ​​then applied to the original side view image to complete the correction from the side view image to the front view image.

[0052] In some specific embodiments, the step of acquiring an image includes:

[0053] S210, installing the image acquisition device and the optical machine on the same bracket and moving them simultaneously.

[0054] S220, using a position sensor device to record the moving distance of the bracket.

[0055] S230. The control system counts the number of pulses returned by the position sensor device, and sends a synchronous refresh signal to the monitoring camera and the optical machine after a certain number of pulses are counted.

[0056] S240. According to the push-scan mechanism of the optical machine, the image is cut in the push-scan direction in units of pixels, and the monitoring camera obtains the real-time exposure image of the optical machine.

[0057] It should be noted here that, while obtaining the exposure pattern under the normal viewing angle, it is necessary to obtain the position of the optical machine relative to the manufacturing plane, and segment the true value of the exposure pattern from the entire slice image according to this relative position to determine whether the current exposure pattern is wrong. The design method of the present invention includes an image acquisition device (visible light camera, infrared camera, etc.) and a position sensing device (grating ruler, encoder, etc.), wherein the position sensing device is used to accurately measure the displacement of the optical machine relative to the manufacturing plane, and the specific method is: the image acquisition device and the optical machine are installed on the same bracket and move at the same time. The position sensing device is used to record the moving distance of the bracket, and the control system counts the number of pulses returned by the position sensing device. After a certain number is counted, a synchronous refresh signal is sent to the monitoring camera and the optical machine. According to the push-sweep mechanism of the optical machine, the image is refreshed in units of a single pixel, and the camera also follows this mechanism when monitoring, and cuts the image in the push-sweep direction in units of pixels. Assume that the sensing coefficient of the grating ruler is y, that is, the number of pulses generated by the displacement sensor for every 1 micron of movement is 1 / y; the actual physical size of a single pixel point of the optical machine exposed along the push-sweep direction is d microns. When the bracket moves the grating ruler to generate d / y pulses, the control system sends a synchronization signal to the monitoring camera and the optical machine. The monitoring camera and the optical machine work at the same time, the optical machine is exposed, and the monitoring camera obtains the real-time exposure image of the optical machine at this time. Since the optical machine and the camera move at the same time, the exposure image of the optical machine and the detection image of the camera are relatively static, and there is no need to worry about the problem of the light source position constantly changing during the exposure of the optical machine.

[0058] In some specific embodiments, the acquired exposure image and the exposure pattern under the normal viewing angle are scaled to the minimum resolution, and a deep learning network is used to determine whether the current exposure pattern is correct.

[0059] In some specific embodiments, the similarity algorithm between frames in S500 is specifically:

[0060] S510, denote the current frame image and the previous frame image as Mi and Mi-1 respectively.

[0061] S520: Scaling and interpolating the current frame image matrix and the previous frame image matrix to reduce the resolution.

[0062] S530: Adjust the current frame image matrix and the previous frame image matrix into a square matrix.

[0063] S540, calculating the difference and eigenvalues ​​of the current frame image matrix and the previous frame image matrix.

[0064] It should be noted here that the similarity algorithm between frames is described as follows:

[0065] Any grayscale image can be described by a matrix. The current frame image and the previous frame image are represented as Mi and Mi-1 respectively. Then, the two image matrices are scaled and interpolated to reduce the resolution while adjusting the matrix to a square matrix for subsequent calculations. The difference between the matrices is calculated, and the calculated difference matrix represents the difference between each element of the two matrices. Then the eigenvalue of the matrix is ​​calculated. The size of the eigenvalue can represent the size of the difference between the two matrices, and it is also the size of the difference between two adjacent frames. The computer can perform parallel calculations on the matrix, and at the same time, it avoids the image correction process in the comparison algorithm. Therefore, this detection speed is very fast and is suitable for long-term and continuous detection application scenarios. After completing the exposure defect detection, the system can feed back the detection results to the device control end, and the device control end performs corresponding compensation and alarm operations based on the detection results to complete the control closed loop of the equipment, ensuring the stability and controllability of the exposure process of large-size ceramic photocuring additive manufacturing.

[0066] Through the parallel image acquisition process and the real-time reception of pulses from the position sensing device to obtain the relative position of the optical machine, when the algorithm completes the correctness detection of one exposure pattern, it can directly continue the detection of the next exposure pattern, so that the computer can adaptively adjust the sampling detection frequency according to its own performance. Taking the current computer with ordinary configuration as an example (processor 12th generation i5, RAM 16G, RTX3050Ti), the detection frame rate can reach more than 80fps, and the adjacent frame detection method is faster, reaching 250-300fps, which is much faster than the 24-30fps of manual vision.

[0067] A system for online monitoring of optical machine exposure patterns based on the same invention includes an image acquisition module for obtaining acquisition images and slice images. A position acquisition module for obtaining the position of the optical machine relative to the manufacturing plane. An image analysis module for analyzing the status of the acquisition image and the slice image. A policy execution module for executing the commands of the image analysis module. It should be noted here that the image acquisition module includes an image sensor, an industrial lens and a narrow-band filter. The position acquisition module includes a position sensor. The image analysis module includes image processing, algorithm analysis and storage modules. The policy execution module includes a device control terminal.

[0068] By adopting the above technical solution disclosed in the present invention, the following beneficial effects are obtained:

[0069] The present invention discloses a system for online monitoring of optical machine exposure graphics and its working method, including the following steps: setting a threshold range of similarity between an acquisition image and a slice image; starting monitoring from the first frame of the acquisition image and the slice image, and using the acquisition image to compare with the slice image to determine the correctness of the exposure; if the first frame is exposed normally, starting from the next frame, directly determine the similarity between the subsequent frame and the previous frame, and if the similarity is within the allowed threshold range, determine that the current exposure is normal; if the similarity between a certain frame and the previous frame exceeds the threshold, it means that the current frame exposure may be wrong, and then several frames continue to use the method of comparing with the slice image to determine whether the exposure is correct; after the detection is stable, continue to use the similarity between frames to determine the correctness of the exposure. The present invention monitors the push-scanning exposure process of ceramic photocuring additive manufacturing throughout the process, and can promptly alarm when abnormalities are found, and is applied to the application scenario of large-size photocuring ceramic additive manufacturing. The frame rate of monitoring is greatly improved.

[0070] After completing the image correction and obtaining the slice true value image, the present invention first scales the two images to a smaller resolution to further speed up the detection speed. Reducing the image resolution does not affect the judgment of the correctness of the exposure. Then, a deep learning network is used to determine whether the current exposure pattern is correct. The specific method is: first collect sample images and corresponding slice images of correct light-machine exposure, non-light-machine exposure, and light-machine exposure pattern errors to form sample pairs of data sets. After collecting enough samples, the sample set is divided into a training set and a validation set, and then the image pairs of the validation set are input into the neural network for training. Finally, the trained model is used to determine whether the light machine is normal. The above-mentioned neural network specifically refers to a twin neural network. The twin neural network uses two identical network branches that share a convolution kernel, which can be used to determine the difference between two images. Each branch of the twin network uses a feature extraction network composed of several convolutional layers, normalization layers, and activation layers to extract the features of the image, output the feature vector of the image, and then perform a differential operation on the feature vectors of the two images, output the differential feature vector of the two images, and input the differential feature vector into a fully connected layer after obtaining it. To avoid data overfitting, a random inactivation layer is added after the fully connected layer, and finally output to the output layer to determine whether the current projection pattern of the optical machine is normal or there is a failure mode such as no light projection / projection pattern error. Using a deep learning network to judge the projection pattern of the optical machine eliminates the step of image threshold segmentation, and GPU parallel computing can be used to accelerate the judgment, which greatly speeds up the detection speed compared to the traditional pixel-by-pixel comparison method.

[0071] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be considered as the scope of protection of the present invention.

Claims

1. A method for online monitoring of light machine exposure patterns, characterized in that: The following steps are involved: S100, setting a similarity threshold range between the collected image and the sliced ​​image; S200, starting monitoring from the first frame of the acquired image and the slice image, and judging the correctness of exposure by comparing the acquired image with the slice image; S300, if the exposure of the first frame of the acquired image is normal, starting from the next frame, directly determine the similarity between the subsequent frame and the previous frame, and if the similarity is within the allowed threshold range, determine that the current exposure is normal; S400, if the similarity between a certain frame and the previous frame exceeds a threshold, it indicates that the current frame may have an exposure error, and subsequent frames are compared with the slice image to determine whether the exposure is correct; S500, after the detection is stable, continue to use the similarity between frames to determine the correctness of exposure; The step of collecting the image comprises: S210, installing the image acquisition device and the optical machine on the same bracket and moving them simultaneously; S220, recording the moving distance of the bracket using a position sensor device; S230, counting the number of pulses returned by the position sensor device through the control system, and sending a synchronous refresh signal to the image acquisition device and the optical machine after a certain number of pulses are counted; S240, according to the push-sweep mechanism of the optical machine, performing image cutting processing in the push-sweep direction in units of pixels, and obtaining a real-time exposure image of the optical machine by the image acquisition device; The optical machine needs to perform several push-scans to complete the exposure of the entire image. During each push-scan process, the position of the optical machine relative to the vertical push-scan direction is a fixed value, and the position in the push-scan direction changes dynamically. The system needs to segment the projection image from the entire slice image according to the relative position of the optical machine, and update the projection image in real time during the movement of the optical machine; The image in the optical machine push-scan process is a continuously changing process, that is, a certain similarity is maintained between frames. By comparing this similarity, it can be determined whether there is an error in image exposure.

2. The method for online monitoring of optical machine exposure patterns according to claim 1, characterized in that: The method of comparing the acquired image with the slice image is specifically as follows: After the monitoring image of the optical system is acquired by the monitoring camera, the image is corrected by using the image processing technology, the side view exposure image is corrected to the front view angle, and the exposure pattern under the front view angle is acquired; According to the position of the optical machine relative to the manufacturing plane, the true value of the exposure pattern is segmented from the entire slice image to determine whether the current exposure pattern is wrong.

3. The method for online monitoring of optical machine exposure patterns according to claim 2, characterized in that: The acquired exposure image and the exposure pattern under the normal viewing angle are scaled to the minimum resolution, and a deep learning network is used to determine whether the current exposure pattern is correct.

4. The method for online monitoring of optical machine exposure patterns according to claim 3, characterized in that: The similarity algorithm between frames in S500 is specifically: S510, representing the current frame image and the previous frame image as Mi and Mi-1 respectively; S520, scaling and interpolating the current frame image matrix and the previous frame image matrix to reduce resolution; S530, adjusting the current frame image matrix and the previous frame image matrix into a square matrix; S540, calculating the difference and eigenvalues ​​of the current frame image matrix and the previous frame image matrix.

5. A system for online monitoring of optical machine exposure patterns, executing the method for online monitoring of optical machine exposure patterns as claimed in any one of claims 1 to 4, characterized in that: include: An image acquisition module, used for obtaining acquisition images and slice images; A position acquisition module, used to acquire the position of the optical machine relative to the manufacturing plane; An image analysis module, used for analyzing the states of the acquired image and the slice image; A policy execution module is used to execute the commands of the image analysis module.

Images