Online defect detection method and system for wiring terminal of optical holographic compressor

Abstract

The invention belongs to the technical field of defect detection, and relates to an online defect detection method and system for an optical holographic compressor wiring terminal, and the method comprises the steps: firstly obtaining a to-be-detected workpiece image through a spatial three-dimensional optical holographic image, and then carrying out the preprocessing of the obtained workpiece image, and obtaining a high-quality, high-definition, low-noise and undistorted image; secondly, inputting the preprocessed image into a defect detection model for detection to obtain a defect detection result, and obtaining information by utilizing a spatial three-dimensional optical holographic image, so that the detection result is clear in visual gradation, rich in stereoscopic impression and high in imaging definition; optical holographic three-dimensional reconstruction of the workpiece is realized, and low time delay, low leak detection and low false detection are designed, so that the project general situation is mastered, the processing historical information of a production line is recorded, and later maintenance of a manufacturer is facilitated.

Description

Technical Field

[0001] The invention belongs to the technical field of defect detection, and relates to an online defect detection method and system for optical holographic compressor terminals. Background Art

[0002] The air compressor is the core component of the air conditioning equipment. The compressor terminals on it can send current into the compressor cover and prevent the high-pressure refrigerant from spilling out. Therefore, whether the compressor terminals can work properly will affect the electrical insulation, air tightness and assemblability of the air conditioning equipment. If the failure of the compressor terminals is not discovered and prevented early, it will lead to a series of tedious picketing work, causing the company to suffer immeasurable economic losses.

[0003] At present, the inspection of compressor terminals mostly adopts machine vision inspection technology. Machine vision inspection is a new non-contact inspection method that has emerged in recent years. It can realize remote measurement and adapt to complex measurement environments such as air and water. It is more flexible than other measurement methods. In addition, industrial cameras have powerful detection capabilities, which makes machine vision inspection high-precision and fast-speed, and the inspection results are more objective, which can better adapt to the development of modern industry. At the same time, machine vision can realize continuous and uninterrupted 24-hour inspection, improving the production efficiency and capacity of enterprises. Due to the above advantages of accuracy, efficiency and repeatability, machine vision inspection has been widely promoted and used in various industrial applications in recent years.

[0004] For example, Liu He et al. proposed a computer image system suitable for automatic detection of apple damage in the document “Research on Automatic Detection of Damage Evaluation Using Computer Image Technology”. The system predicts the location of apple damage by collecting and analyzing the changing images of calyx or stalk; Liu Yunhui et al. from Beijing Jiaotong University proposed a track surface defect detection system in the document “Rail Surface Defect Detection Algorithm Based on Image Processing”, which uses computer image processing technology to automatically detect defects on the surface of railway tracks; Cheng Wansheng from Harbin Institute of Technology used high-speed linear array CCD and large-scale field programmable logic chip (FPGA) technology in the document “Research on Steel Plate Surface Defect Detection Technology” to develop an image preprocessing system. A fully hardware-based high-speed linear array defect collector is used to detect steel plate defects; Wang Lei et al. provided a button battery automatic detection system in the document "Design of button battery automatic detection system based on visual technology", which includes functions such as character correction of battery surface images, and can also realize defect judgment of battery positive and negative surface images; Li Xueqin et al. proposed a magnetic tile defect automatic detection method based on non-subsampled Contourlet domain adaptive threshold surface in the document "Magnetic tile surface defect detection based on non-subsampled Contourlet domain adaptive threshold surface", which uses line scanning to detect NSCT subband coefficients in different regions according to the characteristics of NSCT subband coefficients. The CT coefficients are threshold processed to give normalized adaptive threshold surfaces of different scales and directions; and compared and segmented with the normalized coefficients of each sub-band of the original NSCT to achieve column-by-column adaptive correction of the NSCT coefficients of the magnetic tile image, and finally reconstruct the NSCT coefficients to extract the magnetic tile defects, effectively removing the background texture from the magnetic tile image; Zhang Tengda et al. proposed a defect detection algorithm based on two-dimensional DFT in the document "Surface defect detection of TFT-LCD flat panel based on two-dimensional DFT". The high-energy frequency domain straight line representing the linear texture is detected through Hough transform (HT), and the frequency components in the neighborhood of the straight line are set to 0. The spatial domain image is reconstructed through two-dimensional IDFT to remove the directional linear texture background, and finally a simple The threshold algorithm extracts defects from the background; CN109001215A discloses a machine vision-based power terminal defect detection system, including a conveying platform, a bracket, a lighting device, a collection device, a processing controller and a screening device; a plurality of power terminals are placed on the conveying platform, and the bracket is used to support the lighting device, the collection device and the processing controller; the collection device is used to collect the surface image of each power terminal on the conveyor belt in real time, and then send it to the processing controller; the processing controller processes the surface image and identifies defects, and sends the defect information of the corresponding terminal identified based on the surface image to the screening device; the screening device takes action to screen and pick out the corresponding defective power terminals.

[0005] In the actual application of the detection system at this stage, surface defect detection and classification based on machine vision can greatly improve the efficiency of industrial production. With enough labeled images, the defect detection method based on convolutional neural network has achieved certain detection effects in the existing surface defect detection technology. Although the machine vision image processing algorithm has become increasingly perfect, for some "stubborn" defects, such as texture loss, glass liquid loss, etc., in actual engineering, there will be omissions and false detections in defect detection due to environmental noise and problems with the algorithm itself. For this reason, it is possible to consider extracting the multispectral information of the workpiece as the second entry point to conduct data mining and reduce the omission and false detection rates of small target defects. Moreover, at this stage, it is still difficult to collect defect samples and negative samples of compressor terminal blocks in advance, and the manual marking process is too time-consuming and labor-intensive; the traditional image recognition algorithm has incomplete information in the recognition feature space composed of extracted features under complex scene changes (lighting changes, perspective distortion, occlusion, object deformation, etc.).

[0006] The following shortcomings exist in the appearance defect detection method of compressor terminal products: it is difficult to distinguish black glass beads and defects on glass beads; internal defects such as bubbles and cracks are not easy to observe from the outside; the problem of inconsistent thickness and length of steel needles has a high rate of missed detection and large errors in manual detection. Although relevant research institutions at home and abroad have developed their own defect feature detection systems in many fields, there are currently no reports on the development of appearance defect detection systems for compressor terminal blocks. In particular, considering the complex and changing scenarios of specific industrial environments (lighting changes, perspective distortion, occlusion, object deformation, etc.), the existing detection methods have incomplete information in the identification feature space formed by the extracted features, resulting in invalid fault detection results. Summary of the invention

[0007] The purpose of the present invention is to solve the problems existing in the technology and to provide an optical holographic compressor terminal online defect detection method and system.

[0008] To achieve the above object, the present invention is implemented through the following technical solutions:

[0009] In a first aspect, the present invention provides an optical holographic compressor terminal online defect detection method, which specifically comprises the following steps:

[0010] S1. Acquire the image of the workpiece to be inspected by using a spatial three-dimensional optical holographic image: Make the object light and the reference light meet in space and generate interference through an optical device, and convert the obtained interference pattern into a visible workpiece image;

[0011] S2, preprocessing the acquired workpiece image: performing denoising, contrast enhancement, smoothing, and artifact removal on the workpiece image obtained in step S1 in sequence to obtain a high-quality, clear, low-noise, and distortion-free image;

[0012] S3. Input the preprocessed image into the defect detection model for detection to obtain the defect detection result.

[0013] As a further technical solution of the present invention, the detection process of the defect detection model is:

[0014] S31. Establish a visually reconstructible holographic spectrum perception network;

[0015] S32, using holographic spectrum information capture and perception technology to repair defective areas in the image and obtain a reconstructed sample image;

[0016] S33. Compare the input image and the reconstructed image to indicate the exact defect area.

[0017] As a further technical solution of the present invention, the visually reconstructible holographic spectrum perception network includes:

[0018] Sensors and data acquisition equipment: sensors used to collect optical holographic images and corresponding data acquisition;

[0019] Image preprocessing module: preprocess the collected workpiece image to improve the image quality and clarity;

[0020] Image feature extractor: used to extract features from the preprocessed workpiece image. The feature extraction methods include local feature descriptors and convolutional neural networks (CNNs) to capture the structure, texture, and shape features in the image.

[0021] Spectral perception module: used to perform spectral perception on the extracted image features, and use optical principles or machine learning methods to analyze the spectral information in the image to obtain the spectral characteristics of the object;

[0022] Reconstruction module: Integrate the perceived spectral information with the image features to reconstruct the 3D structure of the object or extract other properties of the object, including image reconstruction using holographic imaging technology or inferring object properties using deep learning methods;

[0023] Application module: A module that applies the reconstruction results to specific tasks. It is designed according to specific application requirements and can include target recognition, target tracking, medical imaging, and virtual reality.

[0024] As a further technical solution of the present invention, the specific process of repairing the defective area in the image is:

[0025] S321, detecting defective areas: detecting the image to determine which areas have defects and the reasons for the defects;

[0026] S322, select a repair method: select an appropriate repair method according to the nature and specific situation of the defect. The repair methods include interpolation, filtering, image fusion, and content-based image filling, which are:

[0027] Interpolation: If the defective area is caused by data loss or pixel damage during acquisition, interpolation methods are used to fill in the missing pixels. Interpolation methods include bilinear interpolation and bicubic interpolation.

[0028] Filtering: If there is noise or image blur in the defect area, use filtering methods to repair it. Filtering includes mean filtering, median filtering, and Gaussian filtering.

[0029] Image fusion: For defective areas with continuous edges, image fusion technology is used to integrate the surrounding pixel information to restore the content of the defective area. Image fusion methods include Laplace pyramid fusion and multi-resolution fusion;

[0030] Content-based image filling: If there is enough information around the defective area, the defective area is repaired using a content-based image filling method, that is, the defective area is filled by searching for similar areas of surrounding pixels based on the texture and structure information of the image;

[0031] S323, adjusting parameters: during the repair process, adjusting the parameters of the repair method to obtain the best repair effect;

[0032] S324, evaluating restoration effect: evaluating the restored image to check whether the restoration effect meets the requirements.

[0033] As a further technical solution of the present invention, the specific process of comparing the input image and the reconstructed image is:

[0034] S331, data preparation: obtaining input image and reconstructed image;

[0035] S332, image alignment: adjusting the input image and the reconstructed image so that the input image and the reconstructed image are spatially aligned so as to accurately compare their pixel values;

[0036] S333, region selection: selecting the region to be compared according to the specific features to be compared;

[0037] S334, comparison method selection: selecting a comparison method for evaluating the difference between the input image and the reconstructed image, the comparison methods include: peak signal-to-noise ratio, used to quantify the difference between the two images, the higher the value, the smaller the difference; structural similarity index, used to evaluate the similarity of image structures, considering brightness, contrast and structure; mean square error, calculating the mean square difference between the pixel values ​​of the two images, the smaller the value, the smaller the difference;

[0038] S335, comparison calculation: using the selected comparison method to calculate the difference between the input image and the reconstructed image;

[0039] S336, result analysis: analyzing the comparison results, evaluating the difference between the input image and the reconstructed image, and determining whether the reconstructed image meets the requirements and whether further improvement or adjustment is needed based on the comparison results;

[0040] S337, Visualization and Reporting: Visualize the comparison results and generate reports to record and share the analysis results. Visualization includes drawing difference maps and histograms to intuitively show the differences between the two images.

[0041] In a second aspect, the present invention provides an optical holographic compressor terminal online defect detection system, comprising:

[0042] An image acquisition block is used to acquire an image of a workpiece to be inspected;

[0043] An image preprocessing module, used for preprocessing the acquired workpiece image;

[0044] The defect detection module is used to detect defects of the workpiece and output defect detection results.

[0045] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in the first aspect is performed.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0047] Compared with the prior art, the present invention has the following advantages:

[0048] (1) Use spatial three-dimensional optical holographic images to obtain information, so that the detection results have clear visual layers, a strong sense of three-dimensionality, and high imaging clarity;

[0049] (2) The optical holographic stereoscopic reconstruction of the workpiece is realized, and the design has low latency, low missed detection and low false detection, so as to grasp the project overview, record the production line processing history information, and facilitate the manufacturer to carry out subsequent maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The present invention is a flowchart of the process of the online defect detection method for optical holographic compressor terminals.

[0051] Figure 2 This is a detection process block diagram of the defect detection model described in the present invention.

[0052] Figure 3 This is a structural block diagram of the visually reconstructible holographic spectrum perception network described in the present invention.

[0053] Figure 4 This is a flowchart of repairing defective areas in an image according to the present invention.

[0054] Figure 5 This is a flowchart of the comparison between the input image and the reconstructed image according to the present invention.

[0055] Figure 6 This is a structural block diagram of the optical holographic compressor terminal online defect detection system described in the present invention. DETAILED DESCRIPTION

[0056] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0057] Embodiment 1:

[0058] like Figure 1 As shown, this embodiment provides an optical holographic compressor terminal online defect detection method, which specifically includes the following steps:

[0059] S1. Acquire the image of the workpiece to be inspected by using a spatial three-dimensional optical holographic image: Make the object light and the reference light meet in space and generate interference through an optical device, and convert the obtained interference pattern into a visible workpiece image;

[0060] S2, preprocessing the acquired workpiece image: performing denoising, contrast enhancement, smoothing, and artifact removal on the workpiece image obtained in step S1 in sequence to obtain a high-quality, clear, low-noise, and distortion-free image;

[0061] S3, input the preprocessed image into the defect detection model for detection, and obtain the defect detection result, such as Figure 2 As shown, the specific detection process is:

[0062] S31. Establish a visually reconstructible holographic spectrum perception network; Figure 3 As shown, the visually reconstructible holographic spectral perception network includes:

[0063] Sensors and data acquisition equipment: sensors used to collect optical holographic images and corresponding data acquisition;

[0064] Image preprocessing module: preprocess the collected workpiece image to improve the image quality and clarity;

[0065] Image feature extractor: used to extract features from the preprocessed workpiece image. The feature extraction methods include local feature descriptors and convolutional neural networks (CNNs) to capture the structure, texture, and shape features in the image.

[0066] Spectral perception module: used to perform spectral perception on the extracted image features, and use optical principles or machine learning methods to analyze the spectral information in the image to obtain the spectral characteristics of the object;

[0067] Reconstruction module: Integrate the perceived spectral information with the image features to reconstruct the 3D structure of the object or extract other properties of the object, including image reconstruction using holographic imaging technology or inferring object properties using deep learning methods;

[0068] Application module: A module that applies the reconstruction results to specific tasks. It is designed according to specific application requirements and can include target recognition, target tracking, medical imaging, and virtual reality.

[0069] S32, using holographic spectrum information capture and perception technology to repair defective areas in the image to obtain a reconstructed image; Figure 4 As shown in the figure, the specific process of repairing the defective area in the image is:

[0070] S321, detecting defective areas: detecting the image to determine which areas have defects and the reasons for the defects. The reasons for the defects include noise, uneven lighting, incomplete recording, etc.;

[0071] S322, select a repair method: select an appropriate repair method according to the nature and specific situation of the defect. The repair methods include interpolation, filtering, image fusion, and content-based image filling, which are:

[0072] Interpolation: If the defective area is caused by data loss or pixel damage during acquisition, interpolation methods are used to fill in the missing pixels. Interpolation methods include bilinear interpolation and bicubic interpolation.

[0073] Filtering: If there is noise or image blur in the defect area, use filtering methods to repair it. Filtering includes mean filtering, median filtering, and Gaussian filtering.

[0074] Image fusion: For defective areas with continuous edges, image fusion technology is used to integrate the surrounding pixel information to restore the content of the defective area. Image fusion methods include Laplace pyramid fusion and multi-resolution fusion;

[0075] Content-based image filling: If there is enough information around the defective area, the defective area is repaired using a content-based image filling method, that is, the defective area is filled by searching for similar areas of surrounding pixels based on the texture and structure information of the image;

[0076] S323, adjusting parameters: During the repair process, adjusting the parameters of the repair method to obtain the best repair effect, for example, adjusting the size of the filter, the parameters of the interpolation method, etc.;

[0077] S324, evaluate the restoration effect: evaluate the restored image to check whether the restoration effect meets the requirements. Quantitative indicators such as PSNR, SSIM, etc. can be used to evaluate the restoration effect, or visual inspection can be performed.

[0078] S33, comparing the input image and the reconstructed image to indicate the exact defect area, such as Figure 5 As shown in Figure 2, the specific process of comparing the input image and the reconstructed image is as follows:

[0079] S331, data preparation: obtaining an input image and a reconstructed image, and saving the input image and the reconstructed image in an appropriate format for subsequent processing and comparison;

[0080] S332, image alignment: adjusting the input image and the reconstructed image so that the input image and the reconstructed image are spatially aligned so as to accurately compare their pixel values;

[0081] S333, region selection: selecting the region to be compared according to the specific features to be compared, and the entire image can be selected for comparison, or only the region of interest can be selected;

[0082] S334, comparison method selection: selecting a comparison method for evaluating the difference between the input image and the reconstructed image, the comparison methods include: PSNR (peak signal-to-noise ratio), used to quantify the difference between the two images, the higher the value, the smaller the difference; SSIM (structural similarity index), used to evaluate the similarity of image structures, considering three aspects: brightness, contrast and structure; MSE (mean square error), calculating the mean square difference between the pixel values ​​of the two images, the smaller the value, the smaller the difference;

[0083] S335, comparison calculation: using the selected comparison method to calculate the difference between the input image and the reconstructed image, which may involve performing difference calculation on pixel values ​​or comparing image structures;

[0084] S336, result analysis: analyzing the comparison results, evaluating the difference between the input image and the reconstructed image, and determining whether the reconstructed image meets the requirements and whether further improvement or adjustment is needed based on the comparison results;

[0085] S337, Visualization Report: Visualize the comparison results and generate a report to record and share the analysis results. Visualization includes drawing difference maps and histograms to intuitively show the differences between the two images.

[0086] Embodiment 2:

[0087] In further embodiments, there is also provided:

[0088] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in Embodiment 1 is performed. For the sake of brevity, it will not be described in detail here.

[0089] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0090] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0091] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in Example 1 is completed.

[0092] The method in Example 1 can be directly embodied as a hardware processor, or a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it is not described in detail here.

[0093] Those skilled in the art will appreciate that the units, i.e., algorithm steps, of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0094] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0095] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. An optical holographic compressor terminal online defect detection method, characterized in that: The specific steps include: S1. Acquire the image of the workpiece to be inspected by using a spatial three-dimensional optical holographic image: Make the object light and the reference light meet in space and generate interference through an optical device, and convert the obtained interference pattern into a visible workpiece image; S2, preprocessing the acquired workpiece image: performing denoising, contrast enhancement, smoothing, and artifact removal on the workpiece image obtained in step S1 in sequence to obtain a high-quality, clear, low-noise, and distortion-free image; S3. Input the preprocessed image into the defect detection model for detection to obtain the defect detection result.

2. The method for online defect detection of optical holographic compressor terminals according to claim 1, characterized in that: The detection process of the defect detection model is as follows: S31. Establish a visually reconstructible holographic spectrum perception network; S32, using holographic spectrum information capture and perception technology to repair defective areas in the image and obtain a reconstructed sample image; S33. Compare the input image and the reconstructed image to indicate the exact defect area.

3. The method for online defect detection of optical holographic compressor terminals according to claim 2, characterized in that: The visually reconstructable holographic spectrum perception network comprises: Sensors and data acquisition equipment: sensors used to collect optical holographic images and corresponding data acquisition; Image preprocessing module: preprocess the collected workpiece image to improve the image quality and clarity; Image feature extractor: used to extract features from the preprocessed workpiece image. The feature extraction methods include local feature descriptors and convolutional neural networks, which are used to capture the structure, texture, and shape features in the image. Spectral perception module: used to perform spectral perception on the extracted image features, and use optical principles or machine learning methods to analyze the spectral information in the image to obtain the spectral characteristics of the object; Reconstruction module: Integrate the perceived spectral information with the image features to reconstruct the three-dimensional structure of the object or extract other attributes of the object, including image reconstruction using holographic imaging technology or inferring object attributes using deep learning methods; Application module: A module that applies the reconstruction results to specific tasks and is designed according to specific application requirements, including target recognition, target tracking, medical imaging, and virtual reality.

4. The method for online defect detection of optical holographic compressor terminals according to claim 2, characterized in that: The specific process of repairing defective areas in an image is as follows: S321, detecting defective areas: detecting the image to determine which areas have defects and the reasons for the defects; S322, select a repair method: select an appropriate repair method according to the nature and specific situation of the defect. The repair methods include interpolation, filtering, image fusion, and content-based image filling, which are: Interpolation: If the defective area is caused by data loss or pixel damage during acquisition, interpolation methods are used to fill in the missing pixels. Interpolation methods include bilinear interpolation and bicubic interpolation. Filtering: If there is noise or image blur in the defect area, use filtering methods to repair it. Filtering includes mean filtering, median filtering, and Gaussian filtering. Image fusion: For defective areas with continuous edges, image fusion technology is used to integrate the surrounding pixel information to restore the content of the defective area. Image fusion methods include Laplace pyramid fusion and multi-resolution fusion; Content-based image filling: If there is enough information around the defective area, the defective area is repaired using the content-based image filling method, that is, the defective area is filled by searching for similar areas of surrounding pixels based on the texture and structure information of the image; S323, adjusting parameters: during the repair process, adjusting the parameters of the repair method to obtain the best repair effect; S324, evaluating restoration effect: evaluating the restored image to check whether the restoration effect meets the requirements.

5. The method for online defect detection of optical holographic compressor terminals according to claim 2, characterized in that: The specific process of comparing the input image and the reconstructed image is: S331, data preparation: obtaining input image and reconstructed image; S332, image alignment: adjusting the input image and the reconstructed image so that the input image and the reconstructed image are spatially aligned so as to accurately compare their pixel values; S333, region selection: selecting the region to be compared according to the specific features to be compared; S334, comparison method selection: selecting a comparison method for evaluating the difference between the input image and the reconstructed image, the comparison methods including: peak signal-to-noise ratio, used to quantify the difference between the two images, the higher the value, the smaller the difference; The structural similarity index is used to evaluate the similarity of image structures, taking into account three aspects: brightness, contrast, and structure; the mean square error calculates the mean square difference between the pixel values ​​of two images, and the smaller the value, the smaller the difference; S335, comparison calculation: using the selected comparison method to calculate the difference between the input image and the reconstructed image; S336, result analysis: analyzing the comparison results, evaluating the difference between the input image and the reconstructed image, and determining whether the reconstructed image meets the requirements and whether further improvement or adjustment is needed based on the comparison results; S337, Visualization and Reporting: Visualize the comparison results and generate reports to record and share the analysis results. Visualization includes drawing difference maps and histograms to intuitively show the differences between the two images.

6. An optical holographic compressor terminal online defect detection system, characterized in that: The method according to any one of claims 1 to 5 is performed, comprising: An image acquisition block is used to acquire an image of a workpiece to be inspected; An image preprocessing module, used for preprocessing the acquired workpiece image; The defect detection module is used to detect defects of the workpiece and output defect detection results.

7. An electronic device, characterized in that: The invention comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 5 is completed.

8. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method described in any one of claims 1 to 5.

Images