Printing fat oil defect detection method based on UNet model and electronic equipment shell

Abstract

The invention relates to the technical field of artificial intelligence, printing and the like, and provides a printing fat oil defect detection method based on a UNet model and an electronic equipment shell, a practical UNet model capable of identifying different fat oil defects is trained, when the electronic equipment shell is printed and produced, an ink printing layer of a current electronic equipment shell base material is imaged, and the printing fat oil defect detection method based on the UNet model is obtained. The method comprises the following steps: obtaining a printing ink printing layer image, inputting the printing ink printing layer image into a practical UNet model, identifying a fat oil defect of a printing pattern of a printing ink printing layer of a current electronic equipment shell base material through the practical UNet model, and when the practical UNet model identifies that the printing ink printing layer of the current electronic equipment shell base material has the fat oil defect of the printing pattern, identifying the current electronic equipment shell base material according to the fat oil defect of the printing pattern of the printing ink printing layer of the current electronic equipment shell base material. And the current electronic equipment shell base material with the fat oil defect is conveyed to a defective product recovery area, so that the detection precision and the detection efficiency are improved when the fat oil defect of the ink printing layer pattern of the base material is detected.

Description

Technical Field

[0001] The present invention relates to the technical fields of artificial intelligence, computer vision, image recognition and printing, and in particular to a detection method for printed fat oil defects based on a UNet model and an electronic device housing. Background Art

[0002] In the printing production process of electronic equipment housing, it generally includes the processes of printing screen preparation, ink printing, standing, baking and laminating. After the printing screen is prepared, the configured ink is printed on the base color layer of the substrate (for example, the green film surface of the substrate) through the printing screen, and then the ink printing layer is covered with a frosted film after standing and baking. In the process of configuring the ink to be printed on the base color layer of the substrate through the printing screen, it is necessary to detect whether there is fat oil defect in the pattern of the ink printing layer of the substrate. In the prior art, in order to detect the fat oil defect of the ink printing layer pattern of the substrate, visual inspection technology is mainly used for inspection. The detection results of visual inspection are easily affected by the operator's experience, vision and changes in ambient light, and the detection accuracy and efficiency are low.

[0003] In summary, in the existing printing production process of electronic device housings, the detection technology for fat oil defects in the ink printing layer pattern of the substrate has technical problems such as low detection accuracy and low detection efficiency. Summary of the invention

[0004] In view of the deficiencies in the above-mentioned prior art, the present invention provides a detection method for printed fat oil defects based on the UNet model and an electronic device housing, so as to improve the detection accuracy and efficiency of fat oil defects in the ink printing layer pattern of the substrate.

[0005] In a first aspect, the present invention provides a method for detecting printing fat oil defects based on a UNet model, comprising: The UNet model is trained by using images of printed patterns containing fat oil defects to obtain a practical UNet model capable of identifying different fat oil defects, wherein the fat oil defects include blurred boundary defects and uneven thickness defects. The blurred boundary defects represent that the ink of the printed pattern diffuses to the area outside the predetermined pattern, resulting in blurred pattern edges. The uneven thickness defects represent that there is too much ink in certain areas of the printed pattern, forming bulges or abnormal thickness. When printing and producing electronic device shells, the ink printing layer of the current electronic device shell substrate after being left to stand and baked is imaged by a diffuse reflection light source camera system to obtain an image of the ink printing layer of the current electronic device shell substrate; Inputting the ink printing layer image of the current electronic device housing substrate into the practical UNet model, and identifying fat oil defects of the printing pattern of the ink printing layer of the current electronic device housing substrate through the practical UNet model; When the practical UNet model identifies that the ink printing layer of the current electronic device housing substrate has a fat oil defect in the printed pattern, the current electronic device housing substrate with the fat oil defect is transported to a defective product recovery area.

[0006] In a second aspect, the present invention provides an electronic device housing, which is detected in a printing process using the above-mentioned UNet model-based printing fat oil defect detection method.

[0007] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a detection method for printed fat oil defects based on a UNet model and an electronic device housing. The UNet model is trained by using an image of a printed pattern containing fat oil defects to obtain a practical UNet model capable of identifying different fat oil defects. When printing and producing the electronic device housing, the ink printing layer of the current electronic device housing substrate after standing and baking is imaged by a diffuse reflection light source camera system to obtain an ink printing layer image of the current electronic device housing substrate. The ink printing layer image of the current electronic device housing substrate is input into the practical UNet model. The fat oil defect of the printed pattern of the ink printing layer of the current electronic device housing substrate is identified by the practical UNet model. When the practical UNet model identifies that the ink printing layer of the current electronic device housing substrate has a fat oil defect in the printed pattern, the current electronic device housing substrate with the fat oil defect is transported to a defective product recovery area, thereby improving the detection accuracy and efficiency when detecting the fat oil defect in the ink printing layer pattern of the substrate. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The drawings described herein are used to provide further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0009] Figure 1 It is a flow chart of a method for detecting printing fat oil defects based on a UNet model according to an embodiment of the present invention.

[0010] Figure 2 It is a flow chart of training a UNet model through an image of a printed pattern containing fat oil defects in an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only an embodiment of a part of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present invention.

[0012] See also Figure 1 , Figure 2 The embodiment of the present invention provides a method for detecting printing fat oil defects based on a UNet model and an electronic device housing. The electronic device housing is detected using the method for detecting printing fat oil defects based on a UNet model in a printing production process. The method for detecting printing fat oil defects based on a UNet model includes: Step S101, training a UNet model with an image of a printed pattern containing fat and oil defects to obtain a practical UNet model capable of identifying different fat and oil defects, wherein the fat and oil defects include a fuzzy boundary defect and an uneven thickness defect, wherein the fuzzy boundary defect indicates that the ink of the printed pattern diffuses to an area outside a predetermined pattern, resulting in a blurred edge of the pattern, and the uneven thickness defect indicates that there is too much ink in certain areas of the printed pattern, forming a bulge or abnormal thickness; Step S102: when printing and producing the electronic device housing, the ink printing layer of the current electronic device housing substrate after being left to stand and baked is imaged by a diffuse reflection light source camera system to obtain an image of the ink printing layer of the current electronic device housing substrate; Step S103, inputting the ink printing layer image of the current electronic device housing substrate into the practical UNet model, and identifying fat oil defects of the printing pattern of the ink printing layer of the current electronic device housing substrate through the practical UNet model; Step S104: When the practical UNet model identifies that the ink printing layer of the current electronic device housing substrate has a fat oil defect in the printed pattern, the current electronic device housing substrate with the fat oil defect is transported to a defective product recovery area.

[0013] It should be noted that in the prior art, in the ink printing of electronic equipment housings, the detection of fat oil defects in printed patterns mainly relies on manual visual inspection. The human eye has limited ability to recognize small or complex defects, and is prone to misjudgment or omission. In addition, in a mass production environment, manual inspection is time-consuming and easy to fatigue, and cannot meet the needs of efficient production. In addition, manual inspection results are affected by the operator's experience, subjective judgment, and changes in ambient light, and lack consistency. In this embodiment, a UNet model is trained by using an image of a printed pattern containing fat oil defects to obtain a practical UNet model that can identify different fat oil defects. When printing and producing an electronic device shell, the ink printing layer of the current electronic device shell substrate after standing and baking is imaged by a diffuse reflection light source camera system to obtain an ink printing layer image of the current electronic device shell substrate. The ink printing layer image of the current electronic device shell substrate is input into the practical UNet model, and the fat oil defects in the printed pattern of the ink printing layer of the current electronic device shell substrate are identified by the practical UNet model. When the practical UNet model identifies that the ink printing layer of the current electronic device shell substrate has fat oil defects in the printed pattern, the current electronic device shell substrate with the fat oil defects is transported to a defective product recovery area, thereby improving the detection accuracy and efficiency when detecting the fat oil defects in the ink printing layer pattern of the substrate.

[0014] It should be noted that the UNet model is widely used in the field of image processing due to its ability to accurately segment image details, and is particularly suitable for processing complex fat defects (such as blurred boundary defects and uneven thickness defects) in the printed pattern in this embodiment. In this embodiment, the UNet model is trained using sample images containing typical fat defects, so that the UNet model can efficiently identify and classify specific types of defects. When the practical UNet model identifies that the ink printing layer of the current electronic device housing substrate has a fat defect in the printed pattern, the current electronic device housing substrate with the fat defect is transported to the defective product recovery area, thereby digitizing and automating the detection process and reducing human intervention and errors.

[0015] It should be noted that in the diffuse reflection light source camera system, the diffuse reflection light source can provide stable lighting conditions, eliminate the impact of light angle and intensity changes on image quality, and ensure imaging consistency. The diffuse reflection light source camera system can delicately capture ink patterns and their defects, providing high-quality image data input for defect recognition of the practical UNet model.

[0016] In some preferred embodiments, when training the UNet model with an image of a printed pattern containing fat oil defects, the method includes: collecting image samples of printed patterns containing fuzzy boundary defects and uneven thickness defects, and annotating the image samples to mark the defective areas and defect types, and obtaining the true labels of the training data; inputting the image samples into the UNet model to train the UNet model, and calculating the loss function and optimizing the UNet model by comparing the difference between the output predicted by the UNet model and the true label, so as to obtain the practical UNet model. It should be noted that in order for the UNet model to effectively learn the characteristics of fat oil defects, image samples containing fuzzy boundary defects and uneven thickness defects are required. These samples represent various defect situations that may be encountered in actual production. By learning these samples, the UNet model can master how to distinguish different types of defects. Each image sample needs to correspond to a true label, and the label marks the defective area and its type. UNet is trained based on the input image samples and their corresponding labels, and can clearly know which areas are defects and what the types of defects are. In this embodiment, by inputting the image sample into the UNet model, the UNet model can extract features (such as edges, textures, etc.) in the image through the convolution layer, and gradually restore these features, and finally predict the defective area in the image. The training of the UNet model is based on the error back propagation mechanism, and the loss function is used to measure the difference between the output predicted by the model and the true label. By calculating this difference, the UNet model can know which areas are not accurately identified, so as to gradually improve the recognition accuracy by optimizing and adjusting its internal parameters. Preferably, the loss function adopts cross entropy loss or Dice coefficient loss. Cross entropy loss or Dice coefficient loss can help the UNet model to be effectively optimized and improve its ability to segment defective areas. In addition, in this embodiment, through multiple training and optimization, the UNet model will gradually improve its ability to recognize fat oil defects, so that the model has high accuracy and robustness in practical applications. The specific optimization process can be to gradually adjust the model parameters (such as convolution kernel weights, network depth, etc.) so that it can better adapt to the needs of the task, for example, in actual production, identify boundary fuzzy defects and uneven thickness defects and perform accurate classification.

[0017] In some preferred embodiments, when the image samples are annotated, the method includes: using a Gaussian filtering algorithm, a median filtering algorithm or a mean filtering algorithm to denoise the image samples to remove random noise in the image samples, and obtaining image samples after random noise removal; using a histogram equalization algorithm or an adaptive histogram equalization algorithm to enhance the contrast of the image samples after random noise removal, so as to highlight the boundary blur defect and the uneven thickness defect, and obtain image samples with highlighted defects. It should be noted that Gaussian filtering is a linear filter that suppresses high-frequency noise (such as random noise) in the image by weighted averaging. It is very effective in removing Gaussian noise and smoothing details in the image. After applying Gaussian filtering, the noise of the image is smoothed, and those randomly distributed interferences can be eliminated, thereby highlighting the real image structure and defect area. Median filtering is a nonlinear filtering method, which is particularly suitable for removing salt and pepper noise (bright spots or dark spots appearing in the image). Compared with Gaussian filtering, median filtering can better maintain the edge features of the image and avoid blurred boundaries. Mean filtering is a simple linear filtering method that performs a local average operation on the image to remove noise in the image. It has a good effect on removing uniformly distributed noise, but may lose some details of the image. Noise in the image may mask or distort the actual defect area, making the annotation inaccurate. Denoising can effectively remove these random noises and retain important image details (such as the outline and texture of the defect), thereby improving the accuracy of subsequent defect area identification. Histogram equalization adjusts the grayscale distribution of the image to make the image contrast more uniform, especially to produce a greater contrast between the dark and light parts of the image, enhance the details of the low-contrast image, and make defects such as blurred boundaries and uneven thickness easier to identify. Unlike traditional histogram equalization, adaptive histogram equalization (CLAHE) performs contrast enhancement in local areas, which is particularly effective for situations where there is a large range of brightness changes in the image. It can effectively avoid the over-enhancement caused by traditional methods and maintain the clarity of image details. In some images, especially in complex printed patterns, the defect area may have a low contrast with the background, making the defect difficult to detect. By improving the contrast of the image, especially highlighting blurred boundaries and uneven thickness defects, the visibility of the defect area can be significantly improved, providing a clear basis for subsequent defect area and type annotation. It should be noted that the boundary fuzziness defect usually manifests as unclear or blurred pattern boundaries. Through denoising and contrast enhancement, the blurred edges can be made more obvious, which is convenient for the subsequent edge detection algorithm to accurately segment. The uneven thickness defect usually causes uneven ink accumulation, forming bulges or abnormal thickness areas. Through image enhancement, the uneven thickness areas can be more easily separated from the background, making it easier to mark these abnormal areas.In general, denoising can eliminate interference factors, and contrast enhancement can highlight the characteristics of defects, making fuzzy boundary defects and uneven thickness defects more obvious, thus facilitating the subsequent annotation process. This process can ensure the accuracy of the annotation results and improve the efficiency and accuracy of annotation.

[0018] In some preferred embodiments, after obtaining an image sample with prominent defects, an edge detection algorithm is used to identify the defect edges of the image in the image sample with prominent defects to obtain a potential defect area with a preliminary outline; the potential defect area is expanded using a region growing algorithm to automatically mark the defect area with blurred boundaries and the defect area with uneven thickness, and the defect type with blurred boundaries and the defect type with uneven thickness are automatically marked according to the regional morphology, edge characteristics and thickness changes of the potential defect area. It should be noted that the main function of edge detection is to identify areas with large changes in light and dark in the image, which usually represent the boundaries or structures in the image. For defects in printed patterns, especially blurred boundary defects and uneven thickness defects, the edges of the defects in the image usually show characteristics different from the background. In this embodiment, the grayscale changes in the image can be captured by an edge detection algorithm (such as Canny edge detection or Sobel operator), and these blurred edges can be identified, thereby highlighting the potential defect area. In addition, uneven thickness defects are usually manifested as excessive accumulation of ink layers in certain areas, resulting in large changes in the thickness of the pattern. Edge detection algorithms can identify the edges of these areas, especially when ink accumulates or is unevenly distributed, the grayscale changes in the image will also show obvious edge features. In addition, the region growing algorithm is a region expansion algorithm based on seed points. It can start from the seed point according to certain similarity criteria (such as grayscale value, texture features) and gradually expand the region until the stop criteria are met. The algorithm can automatically segment the region and mark the corresponding defect area. Fuzzy boundary defects usually have fuzzy edges, and it is not easy to accurately define their range directly using traditional edge detection methods. The region growing algorithm can further expand these edges on the basis of edge detection, so as to accurately mark the complete range of the fuzzy defect area. For uneven thickness defects, the region growing algorithm can identify the areas with uneven ink distribution according to the grayscale distribution or texture features in the image, and expand these areas to finally mark the defective areas. Areas with large thickness changes usually show large fluctuations in grayscale images, and the region growing algorithm can accurately identify these areas. In addition, after marking the potential defect area, it is a crucial step to determine the defect type. Fuzzy boundary defects and uneven thickness defects show different characteristics in the image, so they can be distinguished by regional morphology, edge features and thickness changes. Blurred border defects usually appear as unclear or blurred edges of patterns. After obtaining the preliminary outline through the edge detection algorithm, the edges of the defective area may show a large degree of blurriness, usually with irregular shapes, and the regional morphology is relatively blurred and has no obvious clear boundaries. Uneven thickness defects appear as ink accumulation or abnormal changes in thickness. By analyzing the grayscale value or texture information in the image, areas with too much ink can be identified. Usually, these areas have large grayscale changes and the surface presents more convex features.The process of automatically marking defect types can be based on the analysis of regional morphology (such as regional shape, size, and distribution), edge features (such as edge clarity and smoothness), and thickness changes (such as grayscale distribution and texture features), so as to efficiently identify and mark defective areas, automatically determine the type of defects, and greatly improve marking efficiency and accuracy.

[0019] In some preferred embodiments, the automatically marked defective region with blurred boundaries and the defective region with uneven thickness are compared with the actual defective region to correct the automatically marked defective region with blurred boundaries and the defective region with uneven thickness, and the automatically marked defective type with blurred boundaries and the defective type with uneven thickness are compared with the actual defective type to correct the automatically marked defective type with blurred boundaries and the defective type with uneven thickness, so as to obtain the true label of the training data. It should be noted that although the edge detection and region growing algorithms can effectively mark the defective region, due to image quality, background complexity or algorithm limitations, the automatically marked defective region may have certain errors, such as mislabeling, missing labels or incomplete labels. Comparing the automatically marked defective region with the actual defective region can help identify and correct these errors and ensure the accuracy of the training data. Through this comparison, the inaccurate labeling caused by the limitations of the algorithm can be eliminated to ensure the integrity and accuracy of the defective region. In addition, for defects such as blurred boundaries and uneven thickness, the automatic marking algorithm identifies the defect type through image features (such as edge features, regional morphology and grayscale changes). However, the defect type automatically marked may be affected by background noise, the sensitivity setting of the algorithm, or the complexity of the defect appearance, resulting in misjudgment. In this embodiment, by comparing the defect type automatically marked with the actual defect type, these misjudgments can be identified and corrected. For example, some fuzzy boundaries may be mistakenly marked as uneven thickness, or conversely, some ink layer anomalies may be mistaken for blurred boundaries. By comparing, it can be ensured that the type of defect is correctly identified and adjusted to more accurately conform to the actual situation. Specifically, if there is a difference between the defect type automatically marked and the actual defect type, it can be further adjusted and corrected by analyzing the morphological characteristics of the region (such as shape, symmetry, size, etc.). For example, fuzzy boundary defects often present irregular shapes, while uneven thickness defects usually present local thickness changes. In addition, uneven thickness defects usually show certain texture changes in the image. By comparing with the actual defects, the characteristics of these texture changes can be identified to determine whether they belong to uneven thickness defects. For fuzzy boundary defects, the clarity of the image edge is compared to determine whether it belongs to fuzzy boundary defects. In this embodiment, by comparing and correcting the automatically labeled defect areas and types, the quality of the training data can be effectively improved to ensure that it reflects the actual defect situation. Real labels are crucial for training models because the accuracy and generalization ability of the model are directly affected by the quality of the training data. By correcting the automatically labeled defect areas and defect types, more diverse and high-quality training data can be obtained, which helps the model to be more stable and accurate in identifying different types of defects.

[0020] In some further embodiments, after the current electronic device casing substrate with fat oil defects is transported to the defective product recycling area, if the fat oil defect identified by the practical UNet model is a boundary fuzzy defect, the edge of the fuzzy area is compared with the edge of the normal area to determine the degree of blur; if the fuzzy edge is obvious, it is determined that the boundary fuzzy defect may be caused by uneven printing pressure or ink diffusion, and it is suggested to adjust the printing pressure and printing speed to ensure uniform ink coating, or improve the ink formula to reduce ink diffusion; if the boundary of the fuzzy area is relatively regular, it is determined that the boundary fuzzy defect is caused by too low ink viscosity or too fast solvent evaporation, and it is suggested to increase the proportion of viscous components of the ink to increase the ink viscosity, or improve the ink formula to reduce ink diffusion. Increase the curing speed of the ink to prevent the solvent from evaporating too quickly; if the fat oil defect identified by the practical UNet model is an uneven thickness defect, the distribution of the ink layer in the uneven thickness area is compared to judge the thickness difference; if the ink layer in some areas is obviously too thick, it is determined that the uneven thickness defect may be caused by uneven ink coating or excessive printing pressure, and it is suggested to adjust the coating uniformity of the ink coating equipment, or optimize the printing pressure to ensure uniform ink coating; if the ink layer in some areas is obviously too thin, it is determined that the uneven thickness defect may be caused by insufficient ink supply or too low printing pressure, and it is suggested to increase the ink supply or adjust the printing pressure to ensure uniform ink coating and avoid local ink accumulation or insufficiency. It should be noted that the cause of boundary fuzzy defects is usually related to factors such as ink diffusion or uneven printing pressure. In this embodiment, by comparing the edges of the fuzzy area with the edges of the normal area to judge the degree of fuzziness, different defect types can be effectively distinguished, helping to determine the cause of the defect, thereby accurately adjusting the relevant process parameters or ink formula, reducing subsequent manual inspections and rework, and improving production efficiency. In addition, the uneven thickness defect is mainly manifested as the ink coating being too thick or too thin in a local area. In this embodiment, by comparing the thickness difference and judging the characteristics of the problem area, the specific cause of the defect can be determined, thereby providing an accurate solution.

[0021] In some further embodiments, when the practical UNet model identifies the fat oil defect, it first identifies the fuzzy boundary defect, and then identifies the uneven thickness defect when the fuzzy boundary defect does not exist. It should be noted that the fuzzy boundary defect usually manifests as the blurring or expansion of the edge of the printed pattern, and the area involved is usually the edge of the pattern, with more obvious morphology and characteristics, and in most cases, the blurred edge is usually significantly contrasted with the normal area, and is easily identified by edge detection or image segmentation algorithms. The uneven thickness defect manifests as the thickness difference of the ink layer, which is usually local and more hidden in the image, requiring more accurate detection and identification to determine. When performing fat oil defect detection, this embodiment prioritizes the identification of defect types that are easier to detect, and then identifies defect types that are more difficult to detect. For example, the fuzzy boundary defect is first identified, and when the fuzzy boundary defect does not exist, the uneven thickness defect is identified, thereby improving recognition efficiency and saving computing resources.

[0022] It should be pointed out that the above embodiments are only preferred specific implementation modes of the present invention, and the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention. The protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for detecting printing fat oil defects based on UNet model, characterized in that: include: The UNet model is trained by using images of printed patterns containing fat oil defects to obtain a practical UNet model capable of identifying different fat oil defects, wherein the fat oil defects include blurred boundary defects and uneven thickness defects. The blurred boundary defects represent that the ink of the printed pattern diffuses to the area outside the predetermined pattern, resulting in blurred pattern edges. The uneven thickness defects represent that there is too much ink in certain areas of the printed pattern, forming bulges or abnormal thickness. When printing and producing electronic device shells, the ink printing layer of the current electronic device shell substrate after being left to stand and baked is imaged by a diffuse reflection light source camera system to obtain an image of the ink printing layer of the current electronic device shell substrate; Inputting the ink printing layer image of the current electronic device housing substrate into the practical UNet model, and identifying fat oil defects of the printing pattern of the ink printing layer of the current electronic device housing substrate through the practical UNet model; When the practical UNet model identifies that the ink printing layer of the current electronic device housing substrate has a fat oil defect in the printed pattern, the current electronic device housing substrate with the fat oil defect is transported to a defective product recovery area.

2. The method for detecting printing fat oil defects based on the UNet model according to claim 1, characterized in that: When training a UNet model through an image of a printed pattern containing fat oil defects, the method includes: collecting image samples of a printed pattern containing blurred boundary defects and uneven thickness defects, and annotating the image samples to mark the areas where defects exist and the types of defects, so as to obtain true labels of training data; inputting the image samples into the UNet model to train the UNet model, and by comparing the difference between the output predicted by the UNet model and the true label, calculating the loss function and optimizing the UNet model, so as to obtain the practical UNet model.

3. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 2, characterized in that: The loss function uses cross entropy loss or Dice coefficient loss.

4. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 2, characterized in that: When the image samples are annotated, the steps include: denoising the image samples to remove random noise in the image in the image samples to obtain image samples after random noise removal; contrast enhancing the image samples after random noise removal to highlight the blurred boundary defect and the uneven thickness defect to obtain image samples with highlighted defects.

5. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 4, characterized in that: When denoising the image samples, a Gaussian filtering algorithm, a median filtering algorithm or a mean filtering algorithm is used to denoise the image samples.

6. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 4, characterized in that: When contrast enhancement is performed on the image samples after random noise removal, a histogram equalization algorithm or an adaptive histogram equalization algorithm is used to enhance the contrast of the image samples after random noise removal.

7. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 4, characterized in that: After obtaining the image sample with highlighted defects, identifying the defect edge of the image in the image sample with highlighted defects to obtain a potential defect area with a preliminary outline; The potential defect area is expanded, defect areas with blurred boundaries and defect areas with uneven thickness are automatically marked, and defect types with blurred boundaries and defect types with uneven thickness are automatically marked according to the regional morphology, edge features and thickness changes of the potential defect area.

8. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 7, characterized in that: When identifying the defect edge of the image in the defect-prominent image sample, an edge detection algorithm is used to identify the defect edge of the image in the defect-prominent image sample; when expanding the potential defect area, the potential defect area is expanded using a region growing algorithm.

9. The method for detecting printing fat oil defects based on the UNet model as claimed in claim 7, characterized in that: The defect areas with blurred boundaries and the defect areas with uneven thickness that are automatically marked are compared with the actual defect areas to correct the defect areas with blurred boundaries and the defect areas with uneven thickness that are automatically marked; the defect types with blurred boundaries and the defect types with uneven thickness that are automatically marked are compared with the actual defect types to correct the defect types with blurred boundaries and the defect types with uneven thickness that are automatically marked, so as to obtain the true labels of the training data.

10. An electronic device housing, characterized in that: The electronic device housing is detected during the printing process using the printing fat oil defect detection method based on the UNet model as described in any one of claims 1 to 9.

Images