Ceramic tile finished product defect automatic identification method and device, intelligent terminal and storage medium

Abstract

The application discloses a ceramic tile finished product defect automatic identification method and device, an intelligent terminal and a storage medium. The method comprises the following steps: based on a line scanning industrial camera, an image of a ceramic tile finished product is collected, and the image comprises a low-angle light black-and-white image and a high-angle light black-and-white image; the ceramic tile image is preprocessed; based on the preprocessed ceramic tile image, a defect area corresponding to the low-angle light black-and-white image and the high-angle light black-and-white image is obtained according to Fourier transform; and the defect area is input into a trained double-branch neural network model to obtain an identification result of a defect of the ceramic tile finished product. Compared with the prior art, the accuracy of appearance defect identification of the ceramic tile finished product is high.

Description

Technical Field

[0001] This invention relates to the field of ceramic tile surface defect detection technology, and in particular to an automatic identification method, device, smart terminal and storage medium for finished ceramic tile defects. Background Technology

[0002] With the widespread use of ceramic tiles, the quality requirements for them are becoming increasingly stringent. In actual production, quality inspection is a crucial step in the ceramic tile production line.

[0003] Currently, automated machines are commonly used to inspect the flatness and size of tiles, but surface defects such as cracks, color differences, bumps, dents, holes, and scratches are still mainly detected manually.

[0004] Although there are existing studies on using network models to automatically detect tile appearance defects, the results of existing neural network models in identifying tile defects are not accurate due to the characteristics of tile appearance defects, such as the similarity of defect features to complex tile patterns and the fact that some extremely small defects are only a few pixels.

[0005] Therefore, existing technologies need to be improved and enhanced. Summary of the Invention

[0006] The main objective of this invention is to provide an automatic identification method, device, smart terminal, and storage medium for finished ceramic tile products, which can accurately identify appearance defects in finished ceramic tile products.

[0007] To achieve the above objectives, a first aspect of the present invention provides a method for automatic identification of defects in finished ceramic tiles, the method comprising:

[0008] Based on a line scan industrial camera, images of finished ceramic tiles are acquired, including low-angle illumination black and white images and high-angle illumination black and white images;

[0009] The tile image is preprocessed;

[0010] Based on the preprocessed tile image, the defect areas corresponding to the low-angle illumination black and white image and the high-angle illumination black and white image are obtained according to the Fourier transform.

[0011] The defective area is input into a trained dual-branch neural network model to obtain the identification results of defects in the finished ceramic tile.

[0012] Optionally, during the pre-training of the dual-branch neural network model, the training dataset is also augmented using methods including any one or more of the following: translation, rotation, mirroring, and scaling.

[0013] Optionally, the two branches of the dual-branch neural network model are used to process the defect regions corresponding to the low-angle illumination black-and-white image and the high-angle illumination black-and-white image, respectively. The front end of each branch of the dual-branch neural network model is a Swin transformer model, and the back end is a convolutional neural network model.

[0014] Optionally, the weight parameters of the two Swin transformer models are the same set of weight parameters.

[0015] Optionally, the preprocessing of the tile image includes: performing noise reduction and smoothing processing on the tile image using a bilateral filtering method.

[0016] Optionally, obtaining the defect region based on the Fourier transform includes: obtaining a frequency spectrum image based on the preprocessed tile image using the Fourier transform; obtaining frequency parameters and constructing a filter based on the frequency spectrum image; obtaining a reconstructed image based on the inverse Fourier transform; and performing thresholding processing on the reconstructed image to obtain the defect region.

[0017] Optionally, the identification results of the defects in the finished ceramic tile include any one or more of the following: unevenness defects, size defects, color difference defects, and flatness.

[0018] A second aspect of the present invention provides an automatic defect identification device for finished ceramic tiles, the device comprising:

[0019] An image acquisition unit is used to acquire images of finished ceramic tiles based on a line scan industrial camera. The images include low-angle illumination black and white images and high-angle illumination black and white images.

[0020] A preprocessing unit is used to preprocess the tile image;

[0021] The defect area detection unit is used to obtain the defect areas corresponding to the low-angle illumination black and white image and the high-angle illumination black and white image respectively based on the preprocessed tile image and Fourier transform.

[0022] The defect result extraction unit is used to input the defect area into a trained dual-branch neural network model to obtain the identification result of the defect in the finished tile.

[0023] A third aspect of the present invention provides a smart terminal, the smart terminal including a memory, a processor, and an automatic identification program for finished tile defects stored in the memory and executable on the processor, wherein the automatic identification program for finished tile defects, when executed by the processor, implements any of the steps of the automatic identification method for finished tile defects.

[0024] A fourth aspect of the present invention provides a computer-readable storage medium storing an automatic identification program for defects in finished ceramic tiles, wherein the automatic identification program for defects in finished ceramic tiles, when executed by a processor, implements any of the steps of the automatic identification method for defects in finished ceramic tiles.

[0025] As can be seen from the above, this invention acquires low-angle and high-angle black-and-white images of finished ceramic tiles using a line-scan industrial camera, performs image preprocessing, and obtains defect regions based on Fourier transform. These defect regions are then input into a dual-branch neural network model to identify surface defects in the finished ceramic tiles. Compared with existing technologies, this invention achieves a higher accuracy rate in identifying surface defects in finished ceramic tiles. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a schematic diagram of the automatic identification method for defects in finished ceramic tiles provided in an embodiment of the present invention;

[0028] Figure 2 yes Figure 1 A detailed flowchart of step S300 in the embodiment is shown below;

[0029] Figure 3 This is a schematic diagram of the structure of the automatic defect identification device for finished ceramic tiles provided in an embodiment of the present invention;

[0030] Figure 4 This is a block diagram illustrating the internal structure of a smart terminal provided in an embodiment of the present invention. Detailed Implementation

[0031] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0032] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0033] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0034] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0035] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrases "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0038] With the widespread use of ceramic tiles, the quality requirements for them are becoming increasingly stringent. In actual production, quality inspection is a crucial step in the ceramic tile production line. Currently, automated machines are typically used to inspect the flatness and dimensions of ceramic tiles, but surface defects such as cracks, color differences, bumps, dents, holes, and scratches still primarily rely on manual inspection.

[0039] Although there are existing studies on using network models to automatically detect tile appearance defects, the results of existing neural network models in identifying tile defects are not accurate due to the characteristics of tile appearance defects, such as the similarity of defect features to complex tile patterns and the fact that some extremely small defects are only a few pixels.

[0040] This invention acquires low-angle and high-angle black-and-white images of finished ceramic tiles using a line-scan industrial camera. After image preprocessing, defect regions are obtained using Fourier transform. These defect regions are then input into a dual-branch neural network model to identify surface defects in the finished ceramic tiles. The accuracy of identifying surface defects in finished ceramic tiles is high.

[0041] Exemplary methods

[0042] like Figure 1 As shown, this embodiment of the invention provides an automatic defect identification method for finished ceramic tiles, which can be deployed in a control terminal for inspecting finished ceramic tiles, such as a computer terminal, tablet computer, or smartphone. Specifically, the method includes the following steps:

[0043] Step S100: Based on a line scan industrial camera, acquire images of the finished ceramic tile, including low-angle illumination black and white images and high-angle illumination black and white images;

[0044] Specifically, in this embodiment, the finished ceramic tile to be inspected is conveyed through the inspection area via a conveyor platform. A camera and a laser are respectively mounted at opposite ends of the inspection area via camera and laser brackets, allowing for free adjustment of height and angle. Illumination sources, all square in shape, are provided on both sides of the inspection area.

[0045] Since tile appearance defects are often partial or extremely small defects in large tiles, sometimes only a few pixels, and the defect features are similar to the complex patterns of the tiles, it is difficult to identify the defect features from a tile image obtained from one angle. In addition, some defects can only be captured from specific viewing angles. Therefore, this embodiment simultaneously captures low-angle lighting black and white images and high-angle lighting black and white images, that is, it uses cameras at different heights to capture images of the tiles at the same time, so as to accurately identify tile defects based on images captured from two angles.

[0046] In one embodiment, each camera simultaneously captures three images of each tile to increase recognition accuracy.

[0047] Step S200: Preprocess the tile image;

[0048] Specifically, this embodiment employs a bilateral filtering method to denoise and smooth the tile image. Bilateral filtering is a non-linear filtering method that combines spatial proximity and pixel value similarity in an image, considering both spatial information and grayscale similarity to achieve edge-preserving denoising. It is characterized by its simplicity, non-iterative nature, and locality. The advantage of bilateral filtering is its ability to preserve edges. Traditional Wiener or Gaussian filtering methods, used for denoising, tend to blur edges significantly and do not effectively preserve high-frequency details.

[0049] Optionally, denoising and enhancement of ultrasound images can be performed from both the time and frequency domains. For example, time-domain image denoising and enhancement methods are based on the transformation of the gray values ​​of the image pixels. This includes using various specially designed small sliding windows to filter the local pixel neighborhood of the image, and achieving histogram equalization of the ultrasound image by changing the distribution of pixels at different gray levels. Frequency-domain image enhancement methods mostly use Fourier transform, Z-transform, etc., to transform the time-domain image to the corresponding frequency domain, then selectively adjust and transform a certain frequency or several frequencies, and finally perform an inverse transform back to the time domain.

[0050] Step S300: Based on the preprocessed tile image, obtain the defect areas corresponding to the low-angle illumination black and white image and the high-angle illumination black and white image respectively according to the Fourier transform method;

[0051] Specifically, since defects in tiles account for a very small proportion of the entire tile image, when a neural network model is used to process the entire tile image, the extracted features are often the tile pattern features, while the truly desired defect features are ignored and filtered out. Therefore, in this embodiment, after obtaining the preprocessed tile image, the defect regions corresponding to the low-angle illumination black and white image and the high-angle illumination black and white image are obtained respectively using the Fourier transform method. Thus, the neural network model only focuses on the defect regions, making the extracted features more accurate.

[0052] In one embodiment, such as Figure 2 As shown, the defect region is extracted using the Fourier transform method, specifically including the following steps:

[0053] Step S310: Based on the preprocessed tile image, obtain the frequency spectrum image according to the Fourier transform;

[0054] Step S320: Based on the frequency spectrum image, obtain the frequency parameters and construct a filter for filtering;

[0055] Step S330: Obtain the reconstructed image based on the inverse Fourier transform;

[0056] Step S340: Threshold the reconstructed image to obtain the defect region.

[0057] Specifically, the preprocessed tile image is first subjected to Fourier transform to obtain a frequency spectrum image. Then, the influence of the cutoff frequency parameter on filtering is considered to obtain the frequency parameters of the frequency spectrum image. A filter is constructed based on the frequency parameters for filtering. Then, the reconstructed image is obtained through inverse Fourier transform to achieve the purpose of suppressing background texture information and enhancing defect area information. Finally, the defect area is obtained through binary thresholding and morphological operations (such as edge extraction).

[0058] Step S400: Input the defect area into the trained dual-branch neural network model to obtain the identification result of the defect in the finished tile.

[0059] Specifically, the dual-branch neural network model in this embodiment includes two branches, used to respectively input the defect regions obtained after processing low-angle lighting black-and-white images and high-angle lighting black-and-white images. Each branch of the dual-branch neural network model has the same architecture: a Swin transformer model at the front end and a convolutional neural network model at the back end. The Swin transformer network, as the backbone network for defect region feature extraction, can improve the ability to extract effective image features compared to other network models. Furthermore, the two Swin transformer models at the front end of each branch share the same set of network weight parameters, while the convolutional neural network models at the back end each use independent network weight parameters. The shared network weight parameters enable the two branches to learn feature information from another view, achieving interactive fusion of feature information between different views. Then, by using independent network weight parameters for each convolutional neural network, the output results of each branch are more accurate. Finally, the output features of the two branches are fused and passed through a fully connected layer to obtain the recognition result. This allows for the accurate identification of various defects in finished ceramic tiles, such as unevenness, size defects, color difference defects, and flatness.

[0060] For some large and heavy tiles, the number of samples in the training set may be insufficient when training a two-branch neural network model. To compensate for this lack of training data, this invention first performs image enhancement processing on the tile images in the training dataset. The specific methods of image enhancement are not limited, such as translation, rotation, mirroring, and scaling. Enhancement processing can expand the diversity of the training dataset, increase the number of training data samples, improve model accuracy and generalization ability, and avoid overfitting.

[0061] The specific process of training the dual-branch neural network model is as follows: preprocessing and enhancing the surface defect tile images acquired by the CIS line scan industrial camera, labeling the defects, and dividing them into training set, validation set, and test set according to the proportion; constructing the dual-branch neural network model, and training to obtain the optimized dual-branch neural network model by adjusting the parameters of the training set, validation set, and test set;

[0062] As described above, by preprocessing the low-angle and high-angle illumination black-and-white images of the collected finished tiles and obtaining the defect areas based on Fourier transform, the tile data is segmented, effectively overcoming the problem of being unable to classify scratch defects due to their high similarity to the tile background.

[0063] By inputting the defective area into a dual-branch neural network model, surface defects in finished ceramic tiles can be accurately identified.

[0064] In one embodiment, considering that gradient information in the defect region is more sensitive to defect identification, the Sobel operator (one of the most important operators in pixel image edge detection) is used in the gradient calculation of the dual-branch neural network model.

[0065] During the experimental testing, training was conducted on tiles from actual production. Multiple batches of image data were collected, with each batch containing 1,000 images. The average recognition accuracy reached 95.19%.

[0066] Exemplary device

[0067] like Figure 3 As shown, corresponding to the above-described automatic identification method for defects in finished ceramic tiles, this embodiment of the invention also provides an automatic identification device for defects in finished ceramic tiles, the automatic identification device for defects in finished ceramic tiles comprising:

[0068] Image acquisition unit 600 is used to acquire images of finished ceramic tiles based on a line scan industrial camera;

[0069] Preprocessing unit 610 is used to preprocess the tile image;

[0070] The defect area detection unit 620 is used to obtain the defect area based on the preprocessed tile image and Fourier transform.

[0071] The defect result extraction unit 630 is used to input the defect area into the neural network model to obtain the identification result of the defect in the finished tile.

[0072] Specifically, in this embodiment, the specific functions of each module of the above-mentioned automatic identification device for defects in finished ceramic tiles can be referred to the corresponding description in the above-mentioned automatic identification method for defects in finished ceramic tiles, and will not be repeated here.

[0073] Based on the above embodiments, the present invention also provides a smart terminal, the principle block diagram of which can be as follows: Figure 4As shown. The aforementioned smart terminal includes a processor, memory, network interface, and display screen connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and an automatic defect identification program for finished tiles. The internal memory provides an environment for the operation of the operating system and the automatic defect identification program for finished tiles stored in the non-volatile storage medium. The network interface of the smart terminal is used for communication with external terminals via a network connection. When the automatic defect identification program for finished tiles is executed by the processor, it implements the steps of any of the aforementioned automatic defect identification methods for finished tiles. The display screen of the smart terminal can be a liquid crystal display (LCD) or an e-ink display.

[0074] Those skilled in the art will understand that Figure 4 The block diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the smart terminal to which the present invention is applied. A specific smart terminal may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0075] In one embodiment, a smart terminal is provided, the smart terminal including a memory, a processor, and an automatic tile defect identification program stored in the memory and executable on the processor. When the automatic tile defect identification program is executed by the processor, it performs the following operation instructions:

[0076] Based on a line scan industrial camera, images of finished ceramic tiles are acquired, including low-angle illumination black and white images and high-angle illumination black and white images;

[0077] The tile image is preprocessed;

[0078] Based on the preprocessed tile image, the defect areas corresponding to the low-angle illumination black and white image and the high-angle illumination black and white image are obtained according to the Fourier transform.

[0079] The defective area is input into a trained dual-branch neural network model to obtain the identification results of defects in the finished ceramic tile.

[0080] Optionally, during the pre-training of the dual-branch neural network model, the training dataset is also augmented using methods including any one or more of the following: translation, rotation, mirroring, and scaling.

[0081] Optionally, the two branches of the dual-branch neural network model are used to process the defect regions corresponding to the low-angle illumination black-and-white image and the high-angle illumination black-and-white image, respectively. The front end of each branch of the dual-branch neural network model is a Swin transformer model, and the back end is a convolutional neural network model.

[0082] Optionally, the weight parameters of the two Swin transformer models are the same set of weight parameters.

[0083] Optionally, the preprocessing of the tile image includes: performing noise reduction and smoothing processing on the tile image using a bilateral filtering method.

[0084] Optionally, obtaining the defect region based on the Fourier transform includes: obtaining a frequency spectrum image based on the preprocessed tile image using the Fourier transform; obtaining frequency parameters and constructing a filter based on the frequency spectrum image; obtaining a reconstructed image based on the inverse Fourier transform; and performing thresholding processing on the reconstructed image to obtain the defect region.

[0085] Optionally, the identification results of the defects in the finished ceramic tile include any one or more of the following: unevenness defects, size defects, color difference defects, and flatness.

[0086] This invention also provides a computer-readable storage medium storing an automatic defect identification program for finished ceramic tiles. When executed by a processor, the automatic defect identification program for finished ceramic tiles implements the steps of any of the automatic defect identification methods for finished ceramic tiles provided in this invention.

[0087] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0088] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the above device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0089] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0090] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0091] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of the above modules or units is merely a logical functional division, and in actual implementation, it can be divided in other ways. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0092] If the integrated modules / units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0093] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not mean that the essence of the corresponding technical solutions deviates from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for automatic identification of defects in ceramic tile finished products, characterized by, The method comprises: Based on a line scanning industrial camera, an image of a ceramic tile product is collected, and the image comprises a low-angle light black-and-white image and a high-angle light black-and-white image; Simultaneous collection of images of the ceramic tile using cameras of different heights; Preprocessing of the ceramic tile image; Based on the preprocessed ceramic tile image, a defect area corresponding to the low-angle light black-and-white image and the high-angle light black-and-white image is obtained according to a Fourier transform; The defect area obtained according to the Fourier transform comprises: Based on the preprocessed ceramic tile image, a frequency spectrum image is obtained according to a Fourier transform; Based on the frequency spectrum image, a frequency parameter is obtained and a filter is constructed for filtering; A reconstructed image is obtained according to an inverse Fourier transform; Thresholding processing of the reconstructed image is performed to obtain a defect area; The defect area is input into a trained double-branch neural network model to obtain a recognition result of a defect of the ceramic tile product; The two branches of the double-branch neural network model are used to process the defect area corresponding to the low-angle light black-and-white image and the defect area corresponding to the high-angle light black-and-white image, respectively, and the front end of each branch of the double-branch neural network model is a Swin transformer model, and the back end is a convolutional neural network model; The weight parameters of the two Swin transformer models are the same set of weight parameters; The convolutional neural network models at the back end of the two branches each use independent network weight parameters; The specific process of training the double-branch neural network model comprises: Preprocessing and enhancement of a surface defect ceramic tile image collected by a CIS line scanning industrial camera, defect labeling, and division into a training set, a validation set, and a test set according to a proportion; A double-branch neural network model is constructed, and the model is trained by adjusting the parameters of the training set, the validation set, and the test set to obtain a trained double-branch neural network model; The preprocessing of the ceramic tile image comprises: Denoising and smoothing processing of the ceramic tile image is performed using a bilateral filtering method; And denoising and enhancement processing of the ultrasonic image is performed from both time and frequency domains.

2. The ceramic tile finished product defect automatic recognition method according to claim 1, characterized in that, When pretraining the double-branch neural network model, the training data set is also subjected to enhancement processing, and the enhancement processing method comprises any one or more of the following: Translation, rotation, mirroring, and scaling.

3. The ceramic tile finished product defect automatic recognition method according to claim 1, characterized in that, The recognition result of the defect of the ceramic tile product comprises any one or more of the following: concave-convex defect, size defect, color difference defect, flatness.

4. A device for automatic recognition of defects in ceramic tile finished products, characterized in that, The ceramic tile product defect automatic recognition device is applied to the ceramic tile product defect automatic recognition method of any one of claims 1-3, and the device comprises: An image collection unit configured to collect an image of a ceramic tile product based on a line scanning industrial camera; A preprocessing unit configured to preprocess the ceramic tile image; A defect area detection unit configured to obtain a defect area based on a Fourier transform according to a preprocessed ceramic tile image; A defect result extraction unit configured to input the defect area into a neural network model to obtain a recognition result of a defect of a ceramic tile product.

5. The intelligent terminal, characterized by The intelligent terminal comprises a memory, a processor, and a ceramic tile finished product defect automatic identification program stored in the memory and executable on the processor, and the ceramic tile finished product defect automatic identification program, when executed by the processor, implements the steps of the ceramic tile finished product defect automatic identification method according to any one of claims 1-3.

6. A computer readable storage medium characterized by The computer readable storage medium stores a ceramic tile finished product defect automatic identification program, and the ceramic tile finished product defect automatic identification program, when executed by the processor, implements the steps of the ceramic tile finished product defect automatic identification method according to any one of claims 1-3.

Images