Image data quality detection method and system

Abstract

The invention discloses an image data quality detection method and system, and relates to the technical field of image data detection, and the method comprises the following steps: converting a multi-source read image into a format, taking metadata, checking the integrity, evaluating the definition through edge detection and the like, carrying out the deviation of noise according to a regional standard, comparing colors, carrying out the compensation, then carrying out the judgment, and carrying out the comprehensive rating through fuzzy evaluation. The encrypted image is decrypted firstly, and all the steps are operated according to a specific algorithm and a threshold value. According to the method, the image quality can be accurately detected, multi-dimensional evaluation is comprehensive and accurate, algorithms of all steps are scientific, system modules are high in function and good in collaboration, the detection efficiency and adaptability are improved, and effective application of image data in multiple fields is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of image data detection systems, and in particular to an image data quality detection method and system. Background Art

[0002] In today's digital age, image data is growing explosively and is widely used in many fields, such as medical imaging diagnosis, security monitoring, industrial inspection, digital entertainment, e-commerce, etc. However, the uneven quality of image data has seriously affected its effective application in various fields.

[0003] During the image acquisition process, due to differences in equipment performance, complex and changeable shooting environments, and different technical levels of operators, various image quality problems may occur. For example, low-end image acquisition equipment may produce images with insufficient resolution and large color deviations; images taken in dim light or under strong light are prone to excessive noise, low contrast, and poor clarity; jitter or inaccurate focus during shooting can make the image blurry. In the transmission and storage of images, network failures, damage to storage media, or improper data compression algorithms may cause image data loss, damage, or garbled characters, destroying the integrity of the image.

[0004] In the field of medical imaging, poor image quality may lead to misdiagnosis or missed diagnosis, seriously affecting the treatment effect and life and health of patients. In terms of security monitoring, low-quality images make it difficult to accurately identify key targets such as people and vehicles in the monitoring scene, reducing the reliability and effectiveness of the security system and posing hidden dangers to public safety. In industrial testing, if the image quality cannot be accurately detected, defective products may escape detection and enter the market, damaging the company's reputation and consumer rights. In the field of digital entertainment and e-commerce, image quality directly affects the user experience and product display effect. Blurred and color-distorted images will reduce users' interest in content and desire to buy.

[0005] Existing image quality detection methods are often relatively simple or not comprehensive enough. Some methods can only detect one or several indicators of the image, such as focusing only on clarity or noise level, and lack the ability to comprehensively evaluate multiple aspects such as image integrity and color accuracy. Moreover, some detection methods have poor adaptability to different types of images and application scenarios, and cannot flexibly adjust detection standards and strategies according to the semantic content and purpose of the image. In addition, in terms of the functional integration of the detection system, there is a lack of efficient image reading and format conversion mechanisms, making it difficult to process image data from multiple sources and formats; after discovering image quality problems, there is a lack of effective automatic repair or optimization functions, which cannot meet the needs of image quality improvement and rapid processing in practical applications.

[0006] In summary, with the increasingly extensive and in-depth application of image data, there is an urgent need for a comprehensive, efficient and flexible image data quality detection system and method to ensure the reliable application and value of image data in various fields. Summary of the invention

[0007] The present invention proposes an image data quality detection method and system to solve the problems mentioned in the above-mentioned prior art.

[0008] In order to achieve the above object, the present invention adopts the following technical solution: an image data quality detection method, comprising the following steps: Image reading and format conversion steps: read image data from a variety of storage media or network data sources, and convert the image into a unified internal processing format after reading, including converting color images into a matrix representation based on the YUV color space. During the conversion process, the image metadata is extracted and stored through the image parsing library; Image integrity detection steps: perform integrity check on the converted image, calculate the CRC32 checksum of the image file and compare it with the predefined standard checksum to determine whether data loss or damage occurs during the transmission or storage of the image, check whether the header information of the image is complete and complies with the corresponding format specifications, including file identification, image width and height information, and color depth field, and use a byte-level comparison algorithm. If there is an abnormality, mark the image integrity as damaged; Clarity evaluation steps: Use edge detection-based algorithms to evaluate image clarity, grayscale the image, use the Sobel operator or Canny operator to extract image edges, calculate the number of edge pixels and edge contrast, and based on a pre-set clarity threshold, if the number of edge pixels is lower than the threshold or the edge contrast is insufficient, the image clarity is determined to be substandard. Cluster analysis is also performed on edge pixels to distinguish the clarity contributions of edges of different objects. Noise analysis steps: Use statistical methods to analyze the noise level in the image, divide the image into multiple sub-areas, calculate the standard deviation of the pixel values ​​in each sub-area, and the area with a larger standard deviation indicates that there is more noise. According to the noise distribution and the comparison between the overall noise standard deviation and the preset noise threshold, determine whether the image noise exceeds the range; when calculating the standard deviation, use a weighted calculation method to assign different weights according to the importance of the sub-area's position in the image; Color accuracy judgment steps: for color images, the color accuracy is judged by comparing with a standard color card or a known color reference image, the image is divided into multiple color regions, the main color components and their proportions of each region are extracted, and the color difference is calculated with the corresponding region in the standard color model, including the use of the CIEDE2000 color difference formula. If the calculated color difference exceeds the set color error threshold, it indicates that the image color has deviation. When segmenting the color region, an algorithm based on color space clustering is used to improve the accuracy of color region division, and before the color difference calculation, the image is calibrated in color space to reduce the impact of color deviation caused by equipment differences; Comprehensive quality rating steps: Assign a comprehensive quality rating to the image, set the range of various indicators corresponding to different quality levels, and decide on the subsequent processing of the image based on the level to which the image belongs, including direct use for professional analysis or display or repair or discard. In the rating process, the fuzzy comprehensive evaluation method is used to comprehensively consider the mutual influence of various indicators.

[0009] Furthermore, in the image reading and format conversion steps, if an encrypted image is encountered when reading the image, the corresponding decryption algorithm is used to decrypt it and then convert the format. The decryption algorithm is automatically matched according to the image encryption type. If the image uses the AES encryption algorithm, the system automatically calls the AES decryption module and uses the pre-stored key or the key obtained through the key negotiation protocol for decryption; if it is an RSA encrypted image, the corresponding RSA decryption process is used to ensure that the encrypted image is converted into a processable format.

[0010] Furthermore, in the clarity assessment step, in addition to the edge detection algorithm, a method based on frequency domain analysis is also used, including calculating the Fourier spectrum of the image, using the concentration of the spectrum and the intensity of the high-frequency components to assist in judging the image clarity, using the fast Fourier transform (FFT) algorithm to calculate the image spectrum, and analyzing the proportional relationship between the low-frequency and high-frequency components in the spectrum. If the high-frequency components are small and dispersed, it indicates that the image clarity is low; at the same time, the clarity assessment index is quantified by setting the spectrum threshold.

[0011] Furthermore, in the noise analysis step, different noise tolerance thresholds are used for different types of images, and the noise judgment criteria are flexibly adjusted according to the semantic content and application scenarios of the images. By establishing a mapping database between image types and noise thresholds, the system automatically selects the threshold for noise judgment according to the image type.

[0012] Furthermore, in the color accuracy judgment step, if there are shadows or exposed areas in the image, illumination compensation processing is performed first and then color contrast is performed. An illumination compensation algorithm based on Retinex theory is used to decompose the image into an illumination layer and a reflection layer. The illumination layer is adjusted to eliminate the influence of shadows and highlights, and then color area segmentation and color difference calculation are performed on the compensated image.

[0013] Furthermore, a system for implementing the image data quality detection method includes: Image reading and conversion module: used to implement the functions of image reading and format conversion steps, with compatibility with multiple image formats and metadata extraction capabilities, connecting different storage media and network data sources. When reading images from network data sources, it has network traffic control functions. By setting traffic thresholds and using traffic shaping technology, it can avoid network congestion caused by reading a large number of images. When the network traffic exceeds 80% of the set threshold, it automatically reduces the image reading speed and adjusts the data transmission rate. Integrity detection module: performs image integrity detection steps, ensures image data integrity through checksum calculation and header information inspection, and attempts to automatically repair the image when image integrity is damaged, using data redundancy or error correction code technology to restore part of the lost or damaged data. If a small amount of image data is lost, it is repaired through Reed-Solomon error correction code, and the lost image part is reconstructed based on redundant data information; Clarity assessment module: According to the clarity assessment steps, edge detection or other auxiliary algorithms are used to accurately assess the clarity of the image. It has a clarity enhancement function. For images whose clarity is lower than the threshold, an image sharpening algorithm is used to improve the clarity and then re-evaluate. Specifically, the UnsharpMask sharpening algorithm is used to enhance the image edge contrast and improve the image clarity by adjusting the sharpening parameters, and then the clarity is evaluated again to determine whether it meets the standard; Noise analysis module: responsible for the noise analysis step, using statistical methods combined with different threshold settings to analyze the image noise, generate noise distribution maps, and use heat maps or contour maps to intuitively display the distribution position and intensity of noise in the image. By analyzing the noise distribution map, the noise source or noise concentration area can be located; Color accuracy judgment module: completes the color accuracy judgment step in claim 1, and judges the color of the color image by comparing with the standard and considering special circumstances. This module supports custom color standards, and users can set a reference model for color comparison according to specific needs or industry standards. For example, in the printing industry, users can import a specific printing color standard ICC profile as the basis for color accuracy judgment to ensure that the image color meets the printing requirements; Comprehensive rating module: According to the comprehensive quality rating steps, the comprehensive image quality level is given according to the test results, the quality level classification standards are adjusted according to user needs, and different weight values ​​or threshold ranges are input through the user interface to adapt to the image quality requirements of different application scenarios.

[0014] Compared with the prior art, the present invention has the following beneficial effects: In terms of detection methods, it comprehensively covers steps such as image reading and format conversion, integrity detection, clarity assessment, noise analysis, color accuracy judgment and comprehensive quality rating. Multiple formats of image reading and accurate metadata extraction facilitate traceability management. Integrity detection ensures that data is not damaged or lost. Each evaluation step uses advanced algorithms for different quality dimensions, such as edge detection and frequency domain analysis combined to evaluate clarity, weighted standard deviation calculation and analysis of noise, color comparison after illumination compensation to judge color accuracy, and fuzzy comprehensive evaluation and scientific rating to improve detection accuracy and reliability.

[0015] For the detection system, each module is powerful and works together efficiently. The network traffic control of the image reading and conversion module avoids network congestion, the automatic repair function of the integrity detection module improves image availability, the sharpening and re-evaluation of the clarity evaluation module can save edge images, the noise analysis module's noise map intuitively presents problems, the custom standards of the color accuracy judgment module adapt to diverse needs, and the flexible adjustment of the comprehensive rating module meets different scenarios, which overall improves the comprehensiveness, efficiency and adaptability of image data quality detection, and effectively guarantees the quality and value of image data applications in various fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention is a schematic block diagram of an image data quality detection method proposed by the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0019] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, and it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0020] Reference Figure 1 :A method for detecting image data quality comprises the following steps: Image reading and format conversion steps: Read image data from a variety of storage media or network data sources, supporting common image formats such as JPEG, PNG, BMP, etc. After reading, convert the image into a unified internal processing format, such as converting a color image into a matrix representation based on a specific color space (such as YUV) to facilitate subsequent processing and analysis. During the conversion process, the image metadata (such as image size, resolution, shooting time, etc.) is accurately extracted and stored through a specific image parsing library to ensure the integrity and accuracy of the metadata so that the image source and processing history can be traced later.

[0021] Image integrity detection steps: Perform integrity check on the converted image. By calculating the checksum of the image file (such as CRC32 checksum) and comparing it with the predefined standard checksum, determine whether the image has data loss or damage during transmission or storage. At the same time, carefully check whether the image header information is complete and complies with the corresponding format specifications, including file identification, image width and height information, color depth and other fields. Use byte-level comparison algorithms, and if there is an abnormality, mark the image integrity as damaged.

[0022] Clarity assessment step: Use an edge detection-based algorithm to assess image clarity. First, grayscale the image, then use the Sobel operator or Canny operator to extract the image edge, and calculate the number of edge pixels and edge contrast. According to the pre-set clarity threshold, if the number of edge pixels is lower than the threshold or the edge contrast is insufficient, the image clarity is determined to be substandard. For example, for surveillance images, if the number of edge pixels is less than a specific value and the edge contrast is lower than a certain standard value, key information in the image may not be effectively identified. In this step, cluster analysis can also be performed on edge pixels to distinguish the clarity contributions of the edges of different objects and further refine the clarity assessment results.

[0023] Noise analysis steps: Use statistical methods to analyze the noise level in the image. Divide the image into multiple sub-areas and calculate the standard deviation of the pixel values ​​in each sub-area. Areas with larger standard deviations indicate that there is more noise. Determine whether the image noise exceeds the acceptable range based on the noise distribution and the comparison between the overall noise standard deviation and the preset noise threshold. For medical images or high-precision image detection scenarios, the requirements for noise are more stringent and extremely low noise levels must be maintained. When calculating the standard deviation, a weighted calculation method is used to assign different weights according to the importance of the sub-area's position in the image, so as to more accurately reflect the impact of noise on the overall image quality.

[0024] Color accuracy judgment steps: For color images, the color accuracy is judged by comparing with a standard color card or a known color reference image. The image is divided into multiple color regions, the main color components and their proportions of each region are extracted, and the color difference is calculated with the corresponding region in the standard color model, such as using the CIEDE2000 color difference formula. If the calculated color difference exceeds the set color error threshold, it indicates that the image color is biased, which may affect the use of the image in color-sensitive applications (such as advertising design and artwork reproduction). When segmenting the color region, an algorithm based on color space clustering is used to improve the accuracy of color region division, and before calculating the color difference, the image is calibrated in color space to reduce the impact of color deviation caused by equipment differences.

[0025] Comprehensive quality rating steps: Based on the above-mentioned integrity, clarity, noise, color accuracy and other test results, give the image a comprehensive quality rating. Set the range of indicators corresponding to different quality levels. For example, high-quality images require no abnormalities in integrity, high clarity, low noise and accurate colors; medium-quality images are allowed to have a certain degree of deviation in certain indicators; low-quality images do not reach a good level in multiple indicators. According to the level of the image, the subsequent processing method of the image can be further determined. For example, high-quality images can be directly used for professional analysis or display, while low-quality images may need to be repaired or discarded. In the rating process, the fuzzy comprehensive evaluation method is used to comprehensively consider the mutual influence of various indicators and determine the image quality level more scientifically.

[0026] In the image reading and format conversion steps of the present invention, if an encrypted image is encountered when reading the image, the corresponding decryption algorithm is used to decrypt it and then convert the format. The decryption algorithm is automatically matched according to the image encryption type. If the image adopts the AES encryption algorithm, the system automatically calls the AES decryption module and uses the pre-stored key or the key obtained through the key negotiation protocol for decryption; if it is an RSA encrypted image, the corresponding RSA decryption process is used to ensure that the encrypted image can be smoothly converted into a processable format.

[0027] In the clarity evaluation step of the present invention, in addition to the edge detection algorithm, a method based on frequency domain analysis can also be used, such as calculating the Fourier spectrum of the image, and assisting in judging the image clarity by the concentration of the spectrum and the intensity of the high-frequency component. The image spectrum is calculated using the fast Fourier transform (FFT) algorithm, and the proportional relationship between the low-frequency and high-frequency components in the spectrum is analyzed. If the high-frequency components are small and dispersed, it indicates that the image clarity is low; at the same time, by setting the spectrum threshold, the clarity evaluation index is further quantified to improve the accuracy and objectivity of the evaluation.

[0028] In the noise analysis step, the present invention adopts different noise tolerance thresholds for different types of images (such as natural scenery images, portrait images, and industrial inspection images), and flexibly adjusts the noise judgment standard according to the semantic content and application scenarios of the image. For natural scenery images, due to their rich texture, a relatively high noise standard deviation (such as 10-15) may still be acceptable; while for industrial inspection images, especially high-precision parts inspection, the noise standard deviation needs to be controlled at an extremely low level (such as 2-5). By establishing a mapping database between image types and noise thresholds, the system can automatically select an appropriate threshold for noise judgment according to the image type.

[0029] In the color accuracy judgment step of the present invention, if there are shadows or highlight areas in the image, illumination compensation processing is performed first and then color comparison is performed to improve the reliability of color accuracy judgment. An illumination compensation algorithm based on Retinex theory is used to decompose the image into an illumination layer and a reflection layer, and the illumination layer is adjusted to eliminate the influence of shadows and highlights. Then, color area segmentation and color difference calculation are performed on the compensated image to ensure that the color accuracy evaluation is not disturbed by uneven illumination.

[0030] The present invention also discloses an image data quality detection system, comprising: Image reading and conversion module: It is used to implement the image reading and format conversion steps. It has the compatibility of multiple image formats and metadata extraction capabilities, and can connect to different storage media and network data sources. When reading images from network data sources, it has a network traffic control function. By setting traffic thresholds and using traffic shaping technology, it can avoid network congestion caused by reading a large number of images. For example, when the network traffic exceeds 80% of the set threshold, it automatically reduces the image reading speed and adjusts the data transmission rate to ensure the stable operation of the network.

[0031] Integrity detection module: Perform image integrity detection steps to ensure image data integrity through checksum calculation and header information inspection. When the image integrity is found to be damaged, it can try to automatically repair the image and use data redundancy or error correction code technology to restore some lost or damaged data. If there is a small amount of data loss in the image, it is repaired through Reed-Solomon error correction code, and the lost image part is reconstructed based on the redundant data information to improve the image availability.

[0032] Clarity assessment module: According to the clarity assessment steps, edge detection or other auxiliary algorithms are used to accurately assess the clarity of the image. This module has a clarity enhancement function. For images whose clarity is slightly lower than the threshold, the image sharpening algorithm can be used to improve the clarity and then re-evaluate. For example, the UnsharpMask sharpening algorithm is used to enhance the image edge contrast and improve the image clarity by adjusting the sharpening parameters (such as radius, intensity, etc.), and then the clarity is evaluated again to determine whether it meets the acceptable standard.

[0033] Noise analysis module: responsible for the noise analysis step in claim 1, using statistical methods combined with different threshold settings to analyze image noise. This module can generate a noise distribution map, and use a heat map or contour map to intuitively display the distribution location and intensity of noise in the image, which is convenient for further analysis of image quality issues. By analyzing the noise distribution map, the noise source or noise concentration area can be quickly located, providing a basis for image noise reduction processing.

[0034] Color accuracy judgment module: completes the color accuracy judgment step in claim 1, and judges the color of the color image by comparing with the standard and considering special circumstances. This module supports custom color standards, and users can set a reference model for color comparison according to specific needs or industry standards. For example, in the printing industry, users can import a specific printing color standard ICC profile as the basis for color accuracy judgment to ensure that the image color meets the printing requirements.

[0035] Comprehensive rating module: According to the comprehensive quality rating steps, the comprehensive quality grade of the image is given according to the test results. This module can adjust the quality grade classification standards according to user needs, and input different weight values ​​or threshold ranges through the user interface to adapt to the image quality requirements of different application scenarios. For example, in the field of artistic creation, users can increase the weight of color accuracy so that the comprehensive rating focuses more on the color performance of the image; in the field of security monitoring, the weight of clarity and integrity can be increased to highlight the image's ability to capture key information.

[0036] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for detecting image data quality, characterized in that: The following steps are involved: Image reading and format conversion steps: read image data from a variety of storage media or network data sources, and convert the image into a unified internal processing format after reading, including converting color images into a matrix representation based on the YUV color space. During the conversion process, the image metadata is extracted and stored through the image parsing library; Image integrity detection steps: perform integrity check on the converted image, calculate the CRC32 checksum of the image file and compare it with the predefined standard checksum to determine whether data loss or damage occurs during the transmission or storage of the image, check whether the header information of the image is complete and complies with the corresponding format specifications, including file identification, image width and height information, and color depth field, and use a byte-level comparison algorithm. If there is an abnormality, mark the image integrity as damaged; Clarity evaluation steps: Use edge detection-based algorithms to evaluate image clarity, grayscale the image, use the Sobel operator or Canny operator to extract image edges, calculate the number of edge pixels and edge contrast, and based on a pre-set clarity threshold, if the number of edge pixels is lower than the threshold or the edge contrast is insufficient, the image clarity is determined to be substandard. Cluster analysis is also performed on edge pixels to distinguish the clarity contributions of edges of different objects. Noise analysis steps: Use statistical methods to analyze the noise level in the image, divide the image into multiple sub-areas, calculate the standard deviation of the pixel values ​​in each sub-area, and the area with a larger standard deviation indicates that there is more noise. According to the noise distribution and the comparison between the overall noise standard deviation and the preset noise threshold, determine whether the image noise exceeds the range; when calculating the standard deviation, use a weighted calculation method to assign different weights according to the importance of the sub-area's position in the image; Color accuracy judgment steps: for color images, the color accuracy is judged by comparing with a standard color card or a known color reference image, the image is divided into multiple color regions, the main color components and their proportions of each region are extracted, and the color difference is calculated with the corresponding region in the standard color model, including the use of the CIEDE2000 color difference formula. If the calculated color difference exceeds the set color error threshold, it indicates that the image color has deviation. When segmenting the color region, an algorithm based on color space clustering is used to improve the accuracy of color region division, and before the color difference calculation, the image is calibrated in color space to reduce the impact of color deviation caused by equipment differences; Comprehensive quality rating steps: Assign a comprehensive quality rating to the image, set the range of various indicators corresponding to different quality levels, and decide on the subsequent processing of the image based on the level to which the image belongs, including direct use for professional analysis or display or repair or discard. In the rating process, the fuzzy comprehensive evaluation method is used to comprehensively consider the mutual influence of various indicators.

2. The image data quality detection method according to claim 1, characterized in that: In the image reading and format conversion steps, if an encrypted image is encountered when reading the image, the corresponding decryption algorithm is used to decrypt it and then convert the format. The decryption algorithm is automatically matched according to the image encryption type. If the image uses the AES encryption algorithm, the system automatically calls the AES decryption module and uses the pre-stored key or the key obtained through the key negotiation protocol for decryption; if it is an RSA encrypted image, the corresponding RSA decryption process is used to ensure that the encrypted image is converted into a processable format.

3. The image data quality detection method according to claim 1, characterized in that: In the clarity assessment step, in addition to the edge detection algorithm, a method based on frequency domain analysis is also used, including calculating the Fourier spectrum of the image, using the concentration of the spectrum and the intensity of the high-frequency components to assist in judging the image clarity, using the fast Fourier transform (FFT) algorithm to calculate the image spectrum, and analyzing the proportional relationship between the low-frequency and high-frequency components in the spectrum. If the high-frequency components are small and dispersed, it indicates that the image clarity is low; at the same time, by setting the spectrum threshold, the clarity assessment index is quantified.

4. The image data quality detection method according to claim 1, characterized in that: In the noise analysis step, different noise tolerance thresholds are used for different types of images, and the noise judgment criteria are flexibly adjusted according to the semantic content and application scenarios of the images. By establishing a mapping database between image types and noise thresholds, the system automatically selects the threshold for noise judgment based on the image type.

5. The image data quality detection method according to claim 1, characterized in that: In the color accuracy judgment step, if there are shadows or exposed areas in the image, illumination compensation processing is performed before color contrast. An illumination compensation algorithm based on Retinex theory is used to decompose the image into an illumination layer and a reflection layer. The illumination layer is adjusted to eliminate the influence of shadows and highlights, and then color area segmentation and color difference calculation are performed on the compensated image.

6. A system for implementing the image data quality detection method according to any one of claims 1 to 5, characterized in that: include: Image reading and conversion module: used to implement the functions of image reading and format conversion steps, with compatibility with multiple image formats and metadata extraction capabilities, connecting different storage media and network data sources. When reading images from network data sources, it has network traffic control functions. By setting traffic thresholds and using traffic shaping technology, it can avoid network congestion caused by reading a large number of images. When the network traffic exceeds 80% of the set threshold, it automatically reduces the image reading speed and adjusts the data transmission rate. Integrity detection module: performs image integrity detection steps, ensures image data integrity through checksum calculation and header information inspection, and attempts to automatically repair the image when image integrity is damaged, using data redundancy or error correction code technology to restore part of the lost or damaged data. If a small amount of image data is lost, it is repaired through Reed-Solomon error correction code, and the lost image part is reconstructed based on redundant data information; Clarity assessment module: According to the clarity assessment steps, edge detection or other auxiliary algorithms are used to accurately assess the clarity of the image. It has a clarity enhancement function. For images whose clarity is lower than the threshold, an image sharpening algorithm is used to improve the clarity and then re-evaluate. Specifically, the UnsharpMask sharpening algorithm is used to enhance the image edge contrast and improve the image clarity by adjusting the sharpening parameters, and then the clarity is evaluated again to determine whether it meets the standard; Noise analysis module: responsible for the noise analysis step, using statistical methods combined with different threshold settings to analyze the image noise, generate noise distribution maps, and use heat maps or contour maps to intuitively display the distribution position and intensity of noise in the image. By analyzing the noise distribution map, the noise source or noise concentration area can be located; Color accuracy judgment module: completes the color accuracy judgment step in claim 1, and judges the color of the color image by comparing with the standard and considering special circumstances. This module supports custom color standards, and users can set a reference model for color comparison according to specific needs or industry standards. For example, in the printing industry, users can import a specific printing color standard ICC profile as the basis for color accuracy judgment to ensure that the image color meets the printing requirements; Comprehensive rating module: According to the comprehensive quality rating steps, the comprehensive quality grade of the image is given according to the results of each test, the quality grade classification standards are adjusted according to user needs, and different weight values ​​or threshold ranges are input through the user interface.

Images