Industrial product unmarked tracking method and device based on shearing interference and neural network

Abstract

The invention belongs to the technical field of product tracking, and particularly discloses an industrial product unmarked tracking method and device based on shearing interference and a neural network, and the method comprises the steps: obtaining an unmarked interference image of the surface of an industrial product; preprocessing the interference image, and carrying out binarization processing on the preprocessed interference image to obtain a binarized image; and extracting key features of the binarized image by using a trained neural network model, and matching the key features based on a similarity measurement function to obtain a feature similarity result so as to realize automatic tracking of a product. According to the invention, the tracking efficiency and the identification efficiency of the industrial product are improved.

Description

Technical Field

[0001] The present application belongs to the field of product tracking technology, and more specifically, to a method and device for marking-free tracking of industrial products based on shearing interference and neural network. Background Art

[0002] Current product tracking technologies include QR code tracking technology, barcode tracking technology, quick response code (QRCode), and RFID technology. In general, the technology for product tracking of chips, high-density integrated circuit boards, and ceramic package bases is partially mature. The product tracking technology of chip wafers is mainly completed during the design process, which requires observation under a microscope. The degree of automation is not high and the system cannot be established. Most high-density integrated circuit boards are tracked through RFID technology. The technology and system are mature and marked through electromagnetic waves. The related industrial chain of ceramic package bases is immature, so the tracking technology is not complete, and the overall market demand is large.

[0003] Therefore, the existing technology cannot cope with the tracking of precision products such as chips, high-density integrated circuit boards, and ceramic package bases. It cannot mark the surface of these products, resulting in damage to product performance, and it cannot place RFID tags on the products (some require power, etc.). In addition, the existing technology also has the problem of marking and labeling the inside of the chip and marking the package on the surface. One is that it is extremely cumbersome to observe the inside, and there is a cost for marking the outside. The production process before the chip is packaged cannot be tracked.

[0004] In other words, the tracking technology in the existing technology still has problems such as low degree of automation, immature industrial chain, high labeling cost and complex internal labeling. There are multiple limitations in the tracking of test products, resulting in low efficiency in tracking and identification of test products. Summary of the invention

[0005] In view of the defects of the prior art, the purpose of this application is to provide a markerless tracking method for industrial products based on shearing interferometry and neural networks, aiming to solve the problem of low efficiency in tracking and identifying industrial products.

[0006] In a first aspect, the present application provides a markerless tracking method for industrial products based on shearing interferometry and neural network, comprising:

[0007] Obtaining an interference image without markings on the surface of an industrial product; the interference image is formed by using a beam splitter to split a light beam into two light paths, and after adjusting the optical path difference and shearing amount through a right-angle prism, the two light paths interfere in the overlapping area;

[0008] Preprocessing the interference image, and binarizing the preprocessed interference image to obtain a binarized image;

[0009] The trained neural network model is used to extract the key features of the binary image, and the key features are matched based on a similarity measurement function to obtain feature similarity results to achieve automatic tracking of products.

[0010] The present application utilizes a beam splitter and a right-angle prism to obtain an interference image of an industrial product. The combination of the beam splitter and the right-angle prism allows for more flexible adjustment of the optical path difference and the shear amount, and the optical path can be fine-tuned according to the characteristics of the object being measured to adapt to a more diverse sample surface, such as an irregular or complex surface, thereby improving the versatility and applicability of product tracking. The beam splitter can accurately separate the light beam and, together with the right-angle prism, provide a more stable optical path adjustment, thereby improving the resolution and contrast of the interference fringes, facilitating the accurate detection of minute defects or irregularities on the surface of the product, and improving the detection accuracy and efficiency. Furthermore, by constructing a neural network to pre-process and binarize the interference image, automated image analysis can be achieved, manual intervention can be reduced, work efficiency can be improved, key features of the binary image can be extracted and matched based on a similarity metric function, products can be effectively identified and tracked, and the tracking efficiency and recognition efficiency of industrial products can be improved.

[0011] Optionally, the method for acquiring the interference image includes:

[0012] Collimating the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and controlling the cross-sectional area of ​​the outgoing light beam;

[0013] Split the outgoing light beam into two light paths, and adjust the optical path difference and shearing amount of the two light beams;

[0014] Interference fringes are formed by interfering two light beams in the overlapping area;

[0015] The interference fringes are focused to obtain an interference image. Optionally, the extracting key features of the binary image using a trained neural network model includes:

[0016] The binary image is input into a trained neural network, and the binary image is processed in sequence by an input layer, a convolution layer, a nonlinear activation layer, a pooling layer, a fully connected layer, and an output layer to obtain the key features;

[0017] The neural network is constructed based on the CNN algorithm and trained using image samples and sample labels.

[0018] Optionally, it also includes:

[0019] The SIFT algorithm is used to extract features from the binary image through scale space extreme value detection, key point positioning, direction allocation and key point description to obtain the key features.

[0020] Optionally, preprocessing the interference image and binarizing the preprocessed interference image to obtain a binarized image includes:

[0021] converting the interference image into a grayscale image;

[0022] Performing denoising processing on the grayscale image by using median filtering to remove salt and pepper noise of the grayscale image;

[0023] The preprocessed interference image is binarized to obtain the binarized image.

[0024] Optionally, it also includes:

[0025] When the system is in an offline state, storing the interference image in a database as historical reference data;

[0026] When the system is restored to an online state, the image to be processed is compared with the historical reference data in the database, and it is confirmed that the image to be processed is the interference image.

[0027] This application ensures that the interference images can be effectively stored when the system is offline, and accurate image comparison and confirmation can be performed after the system is restored online, which not only improves the traceability of the data, but also enhances the intelligence level of the system, and can flexibly respond to image processing needs under different conditions. By utilizing historical reference data, the system can better adapt to changes and improve the processing efficiency and accuracy of interference images.

[0028] In a second aspect, the present application also provides a markerless tracking device for industrial products based on shearing interferometry and neural network, comprising:

[0029] An interferometric unit for obtaining an interference image of the surface of industrial products without markings;

[0030] An image processing unit, used for preprocessing the interference image and performing binarization processing on the preprocessed interference image to obtain a binarized image;

[0031] The matching unit is used to extract key features of the binary image using a trained neural network model, match the key features based on a similarity measurement function, and obtain feature similarity results to achieve automatic tracking of products.

[0032] In a third aspect, the present application further provides a shearing interferometry system, comprising:

[0033] Semiconductor laser, beam expansion module, beam splitter, right angle prism and imaging module;

[0034] The semiconductor laser is used to provide a light source with a narrow wavelength band and good coherence;

[0035] The beam expansion module is used to collimate the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and to control the cross-sectional area of ​​the outgoing light beam;

[0036] The beam splitter prism is used to split and shear the outgoing light beam to form interference fringes;

[0037] The right-angle prism is used to adjust the optical path difference and shear amount;

[0038] The imaging module is used to focus the interference fringes to obtain an interference image.

[0039] In a fourth aspect, the present application provides an electronic device comprising: at least one memory for storing programs; and at least one processor for executing the programs stored in the memory. When the programs stored in the memory are executed, the processor is used to execute the method described in the first aspect or any possible implementation of the first aspect.

[0040] In a fifth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method described in the first aspect or any possible implementation of the first aspect.

[0041] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0042] In general, the above technical solutions conceived by this application have the following beneficial effects compared with the prior art:

[0043] (1) The present application utilizes a beam splitter and a right-angle prism to obtain an interference image of an industrial product. The combination of the beam splitter and the right-angle prism allows for more flexible adjustment of the optical path difference and the shear amount. The optical path can be fine-tuned according to the characteristics of the object being measured to adapt to a more diverse sample surface, such as an irregular or complex surface, thereby improving the versatility and applicability of product tracking. The beam splitter can accurately separate the light beam and, together with the right-angle prism, provide a more stable optical path adjustment, thereby improving the resolution and contrast of the interference fringes, facilitating accurate detection of minute defects or irregularities on the surface of the product, and improving detection accuracy and efficiency. Furthermore, by constructing a neural network to pre-process and binarize the interference image, automated image analysis can be achieved, reducing manual intervention and improving work efficiency. By extracting the key features of the binary image and matching them based on a similarity metric function, the product can be effectively identified and tracked, thereby improving the tracking efficiency and recognition efficiency of the industrial product.

[0044] (2) This application realizes automatic identification through AI algorithm and utilizes deep neural network to perform automatic identification, which greatly saves labor costs and improves the level of production automation. The neural network has good learning ability and can be trained and optimized according to the characteristics of different products to meet the detection needs of various products. In addition, through automated image processing and feature matching, it can reduce the errors caused by human operation and improve the accuracy and reliability of detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is one of the flow charts of the method for label-free tracking of industrial products based on shearing interferometry and neural network provided in the embodiment of the present application;

[0046] Figure 2 is a structural schematic diagram of a shearing interferometer system provided in an embodiment of the present application;

[0047] Figure 3 This is the second flow chart of the method for label-free tracking of industrial products based on shearing interferometry and neural network provided in the embodiment of the present application;

[0048] Figure 4 It is a schematic diagram of the structure of a markerless tracking device for industrial products based on shearing interferometry and neural network provided in an embodiment of the present application;

[0049] Figure 5 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0051] The term "and / or" in this article is a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The symbol " / " in this article indicates that the associated objects are in an or relationship, for example, A / B means A or B.

[0052] The terms "first" and "second" in the specification and claims herein are used to distinguish different objects rather than to describe a specific order of the objects. For example, a first response message and a second response message are used to distinguish different response messages rather than to describe a specific order of the response messages.

[0053] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0054] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc.; multiple elements refer to two or more elements, etc.

[0055] Next, the technical solutions provided in the embodiments of the present application are introduced.

[0056] Reference Figure 1 The present application provides a markerless tracking method for industrial products based on shearing interferometry and neural network, comprising:

[0057] S101. Obtaining an interference image of an unmarked surface of an industrial product; the interference image is formed by using a beam splitter to split a light beam into two light paths, and after adjusting the optical path difference and shear amount through a right-angle prism, the two light paths interfere in the overlapping area;

[0058] S102. Preprocessing the interference image, and binarizing the preprocessed interference image to obtain a binary image;

[0059] S103. Use the trained neural network model to extract key features of the binary image, match the key features based on the similarity measurement function, and obtain feature similarity results to achieve automatic tracking of products.

[0060] Specifically, the present application first obtains the interference image of the surface of the product to be tested through a shearing interferometer system. Among them, shearing interferometer is a detection method based on optical interference. A beam of light is divided into two beams, which are respectively irradiated to the object to be tested and the reference surface, and finally the reflected light is recombined to form an interference image. The interference image of the present application can show the optical path difference caused by the slight height change of the product surface, thereby realizing high-precision surface morphology measurement.

[0061] Reference Figure 2 , the shearing interferometer system includes: a semiconductor laser, a beam expansion module, a beam splitter prism, a right-angle prism and an imaging module;

[0062] The semiconductor laser is used to provide a light source with a narrow wavelength band and good coherence;

[0063] The beam expansion module is used to collimate the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and to control the cross-sectional area of ​​the outgoing light beam;

[0064] The beam splitter prism is used to split and shear the outgoing light beam to form interference fringes;

[0065] The right-angle prism is used to adjust the optical path difference and shear amount;

[0066] The imaging module is used to focus the interference fringes to obtain the interference image.

[0067] Specifically, in Figure 2 In the embodiment, the shearing interferometer system includes: a sample to be measured, a semiconductor laser 1, a concave lens 2 and a convex lens 3 of a beam expansion module, a beam splitter prism 4 and a beam splitter prism 5, a right angle prism 6 and a right angle prism 7, and an imaging module 8.

[0068] Specifically, the interference image is obtained according to the above-mentioned shearing interferometry system:

[0069] Collimating the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and controlling the cross-sectional area of ​​the outgoing light beam;

[0070] Split the outgoing light beam into two light paths, and adjust the optical path difference and shearing amount of the two light beams;

[0071] Interference fringes are formed by interfering two light beams in the overlapping area;

[0072] The interference fringes are focused to obtain an interference image. Further, the interference image is preprocessed, and the preprocessed interference image is binarized to obtain a binarized image.

[0073] By inputting the interference image into the pre-trained neural network model, a convolutional neural network (CNN) can be used for image denoising, enhancement and correction. After the preprocessing is completed, the processed image is converted into a binary image using the threshold segmentation method. A binary image refers to an image that contains only two colors (usually black and white). In a binary image, the value of each pixel is only 0 or 1, representing a different color. 0 usually represents black (background) and 1 represents white (foreground or target).

[0074] S3. Use the trained neural network model to extract the key features of the binary image, match the key features based on the similarity measurement function, and obtain feature similarity results to achieve automatic tracking of products.

[0075] The present application obtains key features through feature extraction, which can be done by a traditional SIFT algorithm or a CNN algorithm.

[0076] On the one hand, the SIFT algorithm is used to extract features from the binary image through scale space extreme value detection, key point positioning, direction allocation and key point description to obtain the key features.

[0077] On the other hand, the binary image is input into a trained neural network, and the binary image is processed in sequence by an input layer, a convolutional layer, a nonlinear activation layer, a pooling layer, a fully connected layer, and an output layer to obtain the key features; wherein the neural network is constructed based on a CNN algorithm and trained using image samples and sample labels.

[0078] Specifically, the SIFT (Scale-Invariant Feature Transform) algorithm is a local feature extraction method that has scale invariance and rotation invariance and is often used for image matching and recognition. When extracting features from a binary image, the SIFT process mainly includes the following steps:

[0079] Scale space extremum detection: Find local extreme points in image spaces of different scales as potential feature points.

[0080] Key point positioning: accurately determine the location of feature points, remove edge effects and low-contrast points, and retain stable feature points.

[0081] Direction assignment: According to the gradient information in the neighborhood of the feature point, a main direction is assigned to the feature point to make the feature rotation invariant.

[0082] Key point description: Generate key point descriptors for subsequent feature matching.

[0083] Through the above steps, the SIFT algorithm extracts key features in the binary image and obtains a set of feature points and their descriptors.

[0084] On the other hand, the binary image is input into the pre-trained convolutional neural network (CNN) for processing to extract the key features of deep learning. The specific process includes:

[0085] Input layer: Input the binary image as the starting point of the network.

[0086] Convolution layer: Perform convolution operations on the input image through multiple convolution kernels, extract local features, and generate feature maps.

[0087] Non-linear activation layer: Usually uses activation functions such as ReLU (Rectified Linear Unit) to increase the non-linear ability of the model so that the network can express more complex features.

[0088] Pooling layer: Downsample the feature map output by the convolutional layer (such as maximum pooling) to reduce the resolution of the feature map while retaining important feature information.

[0089] Fully connected layer: Flattens the features extracted by the previous layer and passes through several fully connected neurons to further combine the feature information.

[0090] Output layer: Outputs the processed key features (e.g., the final feature representation for classification or matching) according to the task requirements.

[0091] The CNN network is trained through image samples and corresponding labels. After learning from a large number of image samples, the network can adaptively extract deep features that are useful for image classification or recognition.

[0092] This application matches the extracted features by defining a similarity measurement function. The features of the product to be tested are compared with the pre-stored standard features, and the feature similarity is analyzed. When the match is completed, the system will generate a feature similarity result, and the qualification of the product can be judged according to the set threshold. Based on the matching results, real-time monitoring and automatic tracking are implemented to ensure the consistency and quality of the product during the production process.

[0093] The present application utilizes a beam splitter and a right-angle prism to obtain an interference image of an industrial product. The combination of the beam splitter and the right-angle prism allows for more flexible adjustment of the optical path difference and the shear amount, and the optical path can be fine-tuned according to the characteristics of the object being measured to adapt to a more diverse sample surface, such as an irregular or complex surface, thereby improving the versatility and applicability of product tracking. The beam splitter can accurately separate the light beam and, together with the right-angle prism, provide a more stable optical path adjustment, thereby improving the resolution and contrast of the interference fringes, facilitating the accurate detection of minute defects or irregularities on the surface of the product, and improving the detection accuracy and efficiency. Furthermore, by constructing a neural network to pre-process and binarize the interference image, automated image analysis can be achieved, manual intervention can be reduced, work efficiency can be improved, key features of the binary image can be extracted and matched based on a similarity metric function, products can be effectively identified and tracked, and the tracking efficiency and recognition efficiency of industrial products can be improved.

[0094] Optionally, preprocessing the interference image and binarizing the preprocessed interference image to obtain a binarized image includes:

[0095] converting the interference image into a grayscale image;

[0096] Performing denoising processing on the grayscale image by using median filtering to remove salt and pepper noise of the grayscale image;

[0097] The preprocessed interference image is binarized to obtain the binarized image.

[0098] Specifically, the binarization method in this embodiment adopts the big law, which is an algorithm for automatically determining the global threshold. Its basic idea is to select the best threshold by maximizing the inter-class variance to separate the foreground (such as interference fringes) and the background to the greatest extent. The algorithm assumes that the pixel values ​​of the image are composed of two classes (foreground and background), and the grayscale histogram distributions of these two classes of pixels are separated as much as possible.

[0099] Through the above-mentioned large-law binarization processing, the interference image is converted into a binary image. The image only contains black and white pixel values, which can effectively highlight the structural information of the image and facilitate subsequent feature extraction and analysis.

[0100] That is, the steps of this embodiment are specifically as follows:

[0101] Grayscale: Convert the interference image into a grayscale image to reduce the complexity of image processing.

[0102] Median filtering denoising: Use median filtering to remove salt and pepper noise in grayscale images and maintain the integrity of image details.

[0103] Big Law Binarization: The big law automatically determines the binarization threshold and converts the processed grayscale image into a binary image.

[0104] Optionally, it also includes:

[0105] When the system is in an offline state, storing the interference image in a database as historical reference data;

[0106] When the system is restored to an online state, the image to be processed is compared with the historical reference data in the database, and it is confirmed that the image to be processed is the interference image.

[0107] Specifically, refer to Figure 3 When the system is offline, the interference image of the industrial product will be stored in the database as historical reference data. After the system is restored to the online state, it starts to process the image to be processed and compares the image to be processed with the historical reference data in the database to confirm whether the image to be processed is an interference image.

[0108] Optionally, the comparison process includes feature extraction, similarity calculation, and threshold determination.

[0109] The embodiment of the present application ensures that the interference image can be effectively stored when the system is offline, and accurate image comparison and confirmation can be performed after the system is restored online, which not only improves the traceability of the data, but also enhances the intelligence level of the system, and can flexibly respond to image processing requirements under different states. By using historical reference data, the system can better adapt to changes and improve the processing efficiency and accuracy of interference images.

[0110] Reference Figure 4 The present application also provides a markerless tracking device for industrial products based on shearing interferometry and neural network, comprising:

[0111] Interference unit 410, used to obtain an interference image of the surface of the industrial product without marking; the interference image is formed by using a beam splitter to split the light beam into two light paths, and after adjusting the optical path difference and shearing amount through a right-angle prism, the two light paths interfere in the overlapping area;

[0112] An image processing unit 420 is used to preprocess the interference image and perform binarization processing on the preprocessed interference image to obtain a binarized image;

[0113] The matching unit 430 is used to extract key features of the binary image using the trained neural network model, match the key features based on a similarity measurement function, and obtain feature similarity results to achieve automatic tracking of products.

[0114] Optionally, the method for acquiring the interference image includes:

[0115] Collimating the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and controlling the cross-sectional area of ​​the outgoing light beam;

[0116] Split the outgoing light beam into two light paths, and adjust the optical path difference and shearing amount of the two light beams;

[0117] Interference fringes are formed by interfering two light beams in the overlapping area;

[0118] The interference fringes are focused to obtain an interference image.

[0119] Optionally, the extracting key features of the binary image using a trained neural network model includes:

[0120] The binary image is input into a trained neural network, and the binary image is processed in sequence by an input layer, a convolution layer, a nonlinear activation layer, a pooling layer, a fully connected layer, and an output layer to obtain the key features;

[0121] The neural network is constructed based on the CNN algorithm and trained using image samples and sample labels.

[0122] Optionally, it also includes:

[0123] The SIFT algorithm is used to extract features from the binary image through scale space extreme value detection, key point positioning, direction allocation and key point description to obtain the key features.

[0124] Optionally, preprocessing the interference image and binarizing the preprocessed interference image to obtain a binarized image includes:

[0125] converting the interference image into a grayscale image;

[0126] Performing denoising processing on the grayscale image by using median filtering to remove salt and pepper noise of the grayscale image;

[0127] The preprocessed interference image is binarized to obtain the binarized image.

[0128] Optionally, an offline processing unit is further included, for:

[0129] When the system is in an offline state, storing the interference image in a database as historical reference data;

[0130] When the system is restored to an online state, the image to be processed is compared with the historical reference data in the database, and it is confirmed that the image to be processed is the interference image.

[0131] The present application also provides a shearing interference system, comprising:

[0132] Semiconductor laser, beam expansion module, beam splitter, right angle prism and imaging module;

[0133] The semiconductor laser is used to provide a light source with a narrow wavelength band and good coherence;

[0134] The beam expansion module is used to collimate the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and to control the cross-sectional area of ​​the outgoing light beam;

[0135] The beam splitter prism is used to split and shear the outgoing light beam to form interference fringes;

[0136] The right-angle prism is used to adjust the optical path difference and shear amount;

[0137] The imaging module is used to focus the interference fringes to obtain an interference image. It is understandable that the detailed functional implementation of each of the above units / modules can be found in the introduction of the above method embodiment, and will not be repeated here.

[0138] It should be understood that the above-mentioned device is used to execute the method in the above-mentioned embodiment. The implementation principle and technical effect of the corresponding program module in the device are similar to those described in the above-mentioned method. The working process of the device can refer to the corresponding process in the above-mentioned method, which will not be repeated here.

[0139] Reference Figure 5 Based on the method in the above embodiment, the embodiment of the present application provides an electronic device, which may include: a processor (processor) 510, a communication interface (Communications Interface) 520, a memory (memory) 530 and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call the logic instructions in the memory 530 to execute the method in the above embodiment.

[0140] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application.

[0141] Based on the method in the above embodiment, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method in the above embodiment.

[0142] Based on the method in the above embodiment, an embodiment of the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method in the above embodiment.

[0143] It is understandable that the processor in the embodiment of the present application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor can be a microprocessor or any conventional processor.

[0144] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.

[0145] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions may be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk (SSD)), etc.

[0146] It should be understood that the various numerical numbers involved in the embodiments of the present application are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application.

[0147] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A markerless tracking method for industrial products based on shearing interferometry and neural network, characterized in that: include: Obtaining an interference image without markings on the surface of an industrial product; the interference image is formed by using a beam splitter to split a light beam into two light paths, and after adjusting the optical path difference and shear amount through a right-angle prism, the two light paths interfere in the overlapping area; Preprocessing the interference image, and binarizing the preprocessed interference image to obtain a binarized image; The trained neural network model is used to extract the key features of the binary image, and the key features are matched based on a similarity measurement function to obtain feature similarity results to achieve automatic tracking of products.

2. The method for non-marking tracking of industrial products according to claim 1, characterized in that: The method for acquiring the interference image comprises: Collimating the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and controlling the cross-sectional area of ​​the outgoing light beam; Split the outgoing light beam into two light paths, and adjust the optical path difference and shearing amount of the two light beams; Interference fringes are formed by interfering two light beams in the overlapping area; The interference fringes are focused to obtain an interference image.

3. The method for non-marking tracking of industrial products according to claim 1, characterized in that: The key features of the binary image are extracted using the trained neural network model, including: The binary image is input into a trained neural network, and the binary image is processed by an input layer, a convolution layer, a nonlinear activation layer, a pooling layer, a fully connected layer, and an output layer in sequence to obtain the key features; The neural network is constructed based on the CNN algorithm and trained using image samples and sample labels.

4. The method for non-marking tracking of industrial products according to claim 3, characterized in that: Also includes: The SIFT algorithm is used to extract features from the binary image through scale space extreme value detection, key point positioning, direction allocation and key point description to obtain the key features.

5. The method for tracking industrial products without marking according to claim 1, characterized in that: Preprocessing the interference image and binarizing the preprocessed interference image to obtain a binarized image includes: converting the interference image into a grayscale image; Performing denoising processing on the grayscale image by using median filtering to remove salt and pepper noise of the grayscale image; The preprocessed interference image is binarized to obtain the binarized image.

6. The method for non-marking tracking of industrial products according to claim 1, characterized in that: Also includes: When the system is in an offline state, storing the interference image in a database as historical reference data; When the system is restored to an online state, the image to be processed is compared with the historical reference data in the database, and it is confirmed that the image to be processed is the interference image.

7. A markerless tracking device for industrial products based on shearing interferometry and neural network, characterized in that: include: An interference unit is used to obtain an interference image without markings on the surface of an industrial product; the interference image is formed by using a beam splitter to split the light beam into two light paths, and after adjusting the optical path difference and shearing amount through a right-angle prism, the two light paths interfere in the overlapping area; An image processing unit, used for preprocessing the interference image and performing binarization processing on the preprocessed interference image to obtain a binarized image; The matching unit is used to extract key features of the binary image using a trained neural network model, match the key features based on a similarity measurement function, and obtain feature similarity results to achieve automatic tracking of products.

8. A shearing interferometer system, characterized in that: include: Semiconductor laser, beam expansion module, beam splitter, right angle prism and imaging module; The semiconductor laser is used to provide a light source with a narrow wavelength band and good coherence; The beam expansion module is used to collimate the laser emitted by the semiconductor laser to ensure that the outgoing light beam is parallel light and to control the cross-sectional area of ​​the outgoing light beam; The beam splitter prism is used to split and shear the outgoing light beam to form interference fringes; The right-angle prism is used to adjust the optical path difference and shear amount; The imaging module is used to focus the interference fringes to obtain an interference image.

9. An electronic device, characterized in that: include: at least one memory for storing a computer program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 6.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program runs on a processor, the processor is caused to execute the method according to any one of claims 1 to 6.

Images