Rapid detection method of contaminants on optical component surfaces based on visual imaging

Abstract

The present invention discloses a method for rapid detection of pollutants on the surface of an optical element based on visual imaging, which specifically relates to the technical field of optical element detection, and comprises the following steps: dividing the surface of an optical element into a plurality of detection areas and performing imaging acquisition; analyzing the imaging information of each detection area to generate an environmental change severity index and an image acquisition anomaly index; comprehensively analyzing and generating a detection area change anomaly coefficient; dividing the detection area into an adopted or non-adopted data area according to the anomaly coefficient; analyzing the degree of difference in the environmental change severity index to determine the re-acquisition area; inferring the optimization direction when re-acquiring images; and when re-acquisition is not required, aggregating all images for pollutant detection. The present invention can accurately evaluate the image acquisition quality of each detection area, ensure that the influence of environmental conditions on image acquisition meets the preset standards, thereby improving the overall image data quality, and can avoid unnecessary comprehensive re-acquisition, saving time and resources.

Description

Technical Field

[0001] The present invention relates to the technical field of optical component detection, and more specifically, to a method for quickly detecting contaminants on the surface of an optical component based on visual imaging. Background Art

[0002] Rapid detection of contaminants on the surface of optical components based on visual imaging is a method that uses visual imaging technology to quickly identify and analyze contaminants on the surface of optical components. This technology captures high-resolution images of optical components and uses image processing algorithms to identify the type, size, and distribution of contaminants. This method can provide accurate detection results in a short time, helping to maintain the performance of optical systems and extend their service life.

[0003] The core of this detection method is to use advanced camera equipment and image processing software to non-destructively scan the surface of optical components to generate detailed image data. Then, the image processing algorithm analyzes this data to identify the characteristics of contaminants, such as dust, oil or scratches. Due to its fast, efficient and non-contact characteristics, this detection method is widely used in optical manufacturing, maintenance and quality control, and significantly improves detection efficiency and accuracy.

[0004] However, the basis of rapid detection is high-quality collected image data. The detection process of optical components may be affected by environmental conditions, such as changes in lighting, vibration, and temperature. These factors may lead to unstable image capture, thereby affecting the reliability and consistency of the detection results. How to ensure that the quality of collected image data meets the usage standards is a problem faced by existing technologies. Summary of the invention

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for rapid detection of contaminants on the surface of optical components based on visual imaging comprises the following steps:

[0007] Dividing the surface of the optical element into a plurality of detection areas and performing imaging acquisition on each detection area;

[0008] Analyze the information during the imaging acquisition process of each detection area to generate an environmental change drastic index and an image acquisition abnormality index respectively;

[0009] Based on the comprehensive analysis of the environmental change intensity index and the image acquisition abnormality index, the abnormal change coefficient of the detection area is generated;

[0010] Each detection area is divided into an adopted data area or a non-adopted data area based on the abnormal coefficient of detection area change;

[0011] Perform difference analysis on the environmental change severity index of all detection areas, and judge the area type for re-image acquisition based on the difference analysis results and re-acquire images;

[0012] Based on the results of the environmental change intensity index, image acquisition abnormality index, and difference degree analysis, the optimization direction when re-acquiring images is inferred;

[0013] When there is no need to re-collect images, all collected images are summarized and used as the basis for surface contaminant detection.

[0014] In a preferred embodiment, the environmental change severity index is determined based on the target index in the environmental conditions, and the determination logic of the target index is:

[0015] Data is collected for each preset indicator in the environmental factors, and the correlation coefficient between each indicator in the environmental factors and the image quality is calculated. The calculation formula is:

[0016] ; Indicates the value of variable x, which is the indicator type in the environmental factor. represents the value of variable y, i.e., image quality, represents the average value of the indicator type in the variable x, i.e. the environmental factor. The variable y represents the average value of the image quality. The image quality here is generally set to the image clarity score, and can also be replaced with other image indicators representing the image quality according to the actual use situation, or can be the average value of all image indicators representing the image quality in the present invention. Denotes the correlation coefficient, and then calculates the correlation coefficient The absolute value of , select the indicator corresponding to the largest absolute value as the target indicator.

[0017] In a preferred embodiment, a preset grid division method is used when dividing the surface of the optical element into a plurality of detection areas.

[0018] In a preferred embodiment, the logic for obtaining the environmental change severity index is:

[0019] Collect the target index data of the environmental conditions of each detection area during the imaging acquisition process to obtain the time series P of the target detection area: ; n is the total number of target indicator data, Indicates the target indicator data corresponding to the number n, and then calculates the mean of the time series P to obtain , the formula for calculating the environmental change intensity index is:

[0020] ; Indicates the drastic index of environmental changes in the target detection area. Represents the result of the normalization of the i-th data point, satisfying: ; n is the total number of target indicator data, represents the volatility weight of the ith data point, reflecting the influence of the change amplitude between data points. Indicates the outlier weight of the ith data point, reflecting the impact of the outlier on the VI. The indicator function representing the normalization result of the i-th data point, represents the local standard deviation of the 1st to the i-th data point, represents the target indicator data of the i-th data point, It represents the standard deviation of the time series P. The larger the value, the more drastic the fluctuation of the time series P. Indicates the weight of the standard deviation change rate of the i-th data point, reflecting the impact of the standard deviation change on the VI.

[0021] In a preferred embodiment, a basic threshold is preset. When the absolute value of the result of the data point normalization processing is greater than or equal to the preset basic threshold, the data point is identified as an abnormal data point, and its corresponding target indicator data is an abnormal value. The output result is 1; when the absolute value of the result of the data point standardization is less than the preset basic threshold, the data point is identified as a normal data point, and its corresponding target indicator data is a normal value. The output result is 0.

[0022] In a preferred embodiment, the logic for obtaining the image acquisition abnormality index is:

[0023] First, the image of the target detection area is segmented into multiple regions of interest according to the preset image segmentation strategy (such as K-means clustering, region growing, and graph cutting). Then, anomaly analysis is performed to generate an anomaly report. The missing area occupancy ratio and color abnormal area occupancy ratio of each region of interest, as well as the importance weight of each region of interest are collected from the anomaly report. Then, according to the missing area and color abnormality ratio of each region of interest, the anomaly weight is calculated for each region of interest. The anomaly weight calculation formula is:

[0024] ; represents the anomaly weight of the i-th region of interest, represents the missing area occupancy ratio of the i-th region of interest, represents the proportion of the color abnormal area in the i-th region of interest, , They are all preset non-zero proportional coefficients, which are used to adjust the influence of the missing area occupancy ratio and the color abnormal area occupancy ratio on the abnormal weight;

[0025] Then the relative position weight is calculated for each region of interest. The relative position weight calculation formula is:

[0026] ; represents the relative position weight of the i-th region of interest, It indicates the maximum distance between the center point of all the regions of interest in the target detection area and the preset image center of the optical element. Indicates the distance between the center point of the i-th region of interest in the target detection area and the preset image center of the optical element;

[0027] Finally, the importance weight, relative position weight, and anomaly weight of each region of interest are combined to calculate the image acquisition anomaly index of the target detection area. The calculation formula is:

[0028] ; represents the importance weight of the i-th region of interest, m is the total number of regions of interest divided into the image of the target detection area, Indicates the image acquisition anomaly index of the target detection area.

[0029] In a preferred embodiment, anomaly analysis refers to: constructing a three-dimensional model or a multi-view stitching model of the image based on the collected image data of the target detection area, integrating multi-angle image data, generating an overall view, and then performing image missing detection and color anomaly detection, and then marking the detected image missing and color anomaly areas on the image of the overall view to generate an anomaly report.

[0030] The logic for obtaining the abnormal coefficient of detection area change is as follows: the image acquisition abnormality index and the environmental change drastic index of the target detection area are substituted into the pre-trained machine learning model. The output result of the machine learning model is the abnormal coefficient of detection area change.

[0031] The detection area change anomaly coefficient is compared with the preset change anomaly threshold. If the detection area change anomaly coefficient is greater than or equal to the preset change anomaly threshold, it is divided into a non-adopted data area. If the detection area change anomaly coefficient is less than the preset change anomaly threshold, it is divided into an adopted data area.

[0032] In a preferred embodiment, performing a difference degree analysis on the environmental change severity index of all detection areas refers to: obtaining the environmental change severity index of all detection areas, then calculating the standard deviation of all environmental change severity indexes, and comparing the standard deviation value with a preset fluctuation threshold. If the standard deviation value is greater than or equal to the preset fluctuation threshold, both the adopted data area and the non-adopted data area need to re-capture images. If the standard deviation value is less than the preset fluctuation threshold, only the non-adopted data area needs to re-capture images.

[0033] In a preferred embodiment, reasoning about the optimization direction when re-capturing images refers to:

[0034] The environmental change drastic index, image acquisition anomaly index and difference degree analysis results of the target detection area of ​​the re-captured image are obtained respectively, the environmental change drastic index, image acquisition anomaly index and difference degree analysis results are defined as input variables, and the output variable is defined as the direction to be optimized. The input variable is fuzzified and the value of the input variable is converted into a fuzzy set. The output variable is fuzzified and the output variable is converted into a fuzzy set. Fuzzy rules are formulated to describe the optimization requirements under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to determine the direction to be optimized.

[0035] Technical effects and advantages of the present invention:

[0036] The present invention generates the detection area change anomaly coefficient through comprehensive analysis of the environmental change drastic index and the image acquisition anomaly index, and accurately evaluates the image acquisition quality of each detection area. The environmental change drastic index of each detection area is evaluated by difference degree analysis to ensure that the impact of environmental conditions on image acquisition meets the preset standards, thereby improving the overall image data quality.

[0037] Based on the difference degree analysis results, the present invention reasonably determines whether it is necessary to re-collect images of all detection areas, avoiding unnecessary comprehensive re-collection and saving time and resources. The optimization direction when re-collecting images is inferred, and fuzzy logic is used to optimize image collection and environmental control strategies to improve resource utilization efficiency.

[0038] The present invention calculates the environmental change intensity index in real time, dynamically adjusts the detection parameters, i.e., the detection environment parameters and acquisition strategies, enhances the system's adaptability to environmental changes, and uses the standard deviation analysis method to identify and respond to the impact of environmental changes, ensuring that the system can still operate stably under complex environmental conditions. The preset grid division method is used to accurately divide the surface of the optical element, improve the parallelism of image acquisition and processing, and speed up the detection process. Automated image processing and analysis technology reduces manual intervention and improves overall detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings;

[0040] Figure 1 This is a schematic diagram of the principle of the method for rapid detection of contaminants on the surface of optical components based on visual imaging in the present invention. DETAILED DESCRIPTION

[0041] 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.

[0042] Reference Figure 1 The following embodiments are obtained:

[0043] Embodiment 1: A method for rapidly detecting contaminants on the surface of an optical component based on visual imaging. The method for rapidly detecting contaminants on the surface of an optical component based on visual imaging comprises the following steps:

[0044] Dividing the surface of the optical element into a plurality of detection areas and performing imaging acquisition on each detection area;

[0045] Analyze the information during the imaging acquisition process of each detection area to generate an environmental change drastic index and an image acquisition abnormality index respectively;

[0046] Based on the comprehensive analysis of the environmental change intensity index and the image acquisition abnormality index, the abnormal change coefficient of the detection area is generated;

[0047] Each detection area is divided into an adopted data area or a non-adopted data area based on the abnormal coefficient of detection area change;

[0048] Perform difference analysis on the environmental change severity index of all detection areas, and judge the area type for re-image acquisition based on the difference analysis results and re-acquire images;

[0049] Based on the results of the environmental change intensity index, image acquisition abnormality index, and difference degree analysis, the optimization direction when re-acquiring images is inferred;

[0050] When there is no need to re-collect images, all collected images are summarized and used as the basis for surface contaminant detection.

[0051] The environmental change severity index measures the degree of influence of various environmental factors on image quality based on the determination of the target indicators in the environmental conditions. By calculating the correlation coefficient between each indicator in the environmental factors and the image quality, it can be identified which environmental factor has the most significant impact on the image quality. The logic for determining the target indicator is: data is collected for each preset indicator in the environmental factors, and the correlation coefficient between each indicator in the environmental factors and the image quality is calculated. The calculation formula is:

[0052] ; Indicates the value of variable x, which is the indicator type in the environmental factor. represents the value of variable y, i.e., image quality, represents the average value of the indicator type in the variable x, i.e. the environmental factor. The variable y represents the average value of the image quality. The image quality here is generally set to the image clarity score, and can also be replaced with other image indicators representing the image quality according to the actual use situation, or can be the average value of all image indicators representing the image quality in the present invention. Denotes the correlation coefficient, and then calculates the correlation coefficient The absolute value of the maximum absolute value is selected as the target indicator; for example, within a fixed time window, the environmental data of each detection area is collected, including light, vibration, and temperature, etc. The image quality data of each detection area is collected, including image clarity score, graphic edge strength score, etc. The correlation coefficient The larger the absolute value of , the stronger the linear relationship between the environmental factor and the image quality. The environmental factor with the largest absolute value of the correlation coefficient is selected as the most significant environmental factor.

[0053] When dividing the surface of the optical element into multiple inspection areas, use the preset grid division method to determine the size and shape of the optical element: measure the actual size (length, width, height) and shape (such as round, square, rectangular, etc.) of the optical element. Purpose: Ensure that the grid division can cover the entire surface of the optical element. Select the size and resolution of the grid: Determine the resolution of the grid, that is, the size of each grid unit. It is usually determined according to the inspection requirements and the characteristics of the surface of the optical element. When selecting the grid size, it is necessary to consider the balance between inspection accuracy and efficiency. Purpose: Determine the size of each grid unit for easy inspection and analysis. According to the size of the optical element and the preset grid size, calculate the number of grid units to be divided. Use mathematical methods to generate a regular grid on the surface of the optical element. For example, assuming that the length of the optical element is L, the width is W, and the side length of each grid unit is d, then the number of rows R of the grid is the ratio of the length L to the side length d rounded up, and the number of columns C is the ratio of the width W to the side length d rounded up. The generated grid units are numbered and marked to facilitate subsequent data recording and analysis. Use the row and column numbering method, for example, the first row and first column is (1,1), the first row and second column is (1,2), and so on. Purpose: To ensure that each grid cell has a unique identifier to facilitate subsequent data processing and analysis.

[0054] The logic for obtaining the environmental change drastic index is:

[0055] Collect the target index data of the environmental conditions of each detection area during the imaging acquisition process to obtain the time series P of the target detection area: ; n is the total number of target indicator data, Indicates the target indicator data corresponding to the number n, and then calculates the mean of the time series P to obtain , the formula for calculating the environmental change intensity index is:

[0056] ; It represents the result of the normalization of the ith data point, reflecting the degree of deviation of each data point from the mean, satisfying: ; n is the total number of target indicator data, represents the volatility weight of the ith data point, reflecting the influence of the change amplitude between data points. Indicates the outlier weight of the ith data point, reflecting the impact of the outlier on the VI. The indicator function representing the normalization result of the i-th data point, represents the local standard deviation of the 1st to the i-th data point, represents the target indicator data of the i-th data point, It represents the standard deviation of the time series P. The larger the value, the more drastic the fluctuation of the time series P. The standard deviation change rate weight of the i-th data point reflects the impact of the standard deviation change on the VI; the change amplitude term Represents the absolute change between adjacent data points and is multiplied by the change weight , the amplitude term reflects the amplitude of the change between data points, and the standard deviation change term It represents the local standard deviation change rate of adjacent data points and is multiplied by the standard deviation change rate weight. The standard deviation change term reflects the degree of change in data stability. Indicates the drastic index of environmental changes in the target detection area. The larger the value, the more drastic the environmental change and the greater the impact on the image quality. The preset weight coefficient mentioned in the present invention can be a preset based on experience and professional knowledge, or it can be determined through experiments and data analysis, or it can be a multi-factor analysis method (such as principal component analysis, factor analysis) to evaluate the contribution of each factor or other methods mentioned in the prior art, which will not be repeated here.

[0057] A basic threshold is preset. When the absolute value of the result of the data point normalization processing is greater than or equal to the preset basic threshold, the data point is identified as an abnormal data point, and its corresponding target indicator data is an abnormal value. The output result is 1; when the absolute value of the result of the data point standardization is less than the preset basic threshold, the data point is identified as a normal data point, and its corresponding target indicator data is a normal value. The output result of is 0; outliers usually represent extreme situations or special events, which have a significant impact on image quality. By calculating the outlier items, these anomalies can be identified and their impact can be weighed. When the absolute value of the standardized result of a data point is greater than the preset threshold, it is marked as an outlier and assigned a corresponding weight to ensure that the anomaly is fully considered in the calculation of the environmental change drastic index. Outliers may have a significant impact on the detection results. By identifying and weighting outliers, the detection accuracy can be improved to ensure that the environmental change drastic index more accurately reflects the actual situation. By calculating the outlier items, the impact of abnormal data points on the environmental change drastic index is quantified, so that it occupies the appropriate weight in the overall evaluation. By quantifying the impact of environmental changes, the stability and reliability of the system can be enhanced, so that the system can maintain a high detection performance under different environmental conditions. Using the environmental change drastic index as a system monitoring indicator, the impact of environmental changes on system performance is evaluated in real time to ensure that the system can still operate stably in a complex environment. By calculating the environmental change drastic index, the detection process can be optimized, the errors and deviations caused by environmental changes can be reduced, and the detection efficiency and accuracy can be improved. During the detection process, the environmental change drastic index value is calculated in real time, and the detection parameters and processes are optimized to ensure the reliability and consistency of the detection results, thereby serving as the final image basis for rapid detection of contaminants on the surface of optical components.

[0058] The logic for obtaining the image acquisition anomaly index is:

[0059] First, the image of the target detection area is segmented into multiple regions of interest according to the preset image segmentation strategy (such as K-means clustering, region growing, and graph cutting). Then, anomaly analysis is performed to generate an anomaly report. The missing area occupancy ratio and color abnormal area occupancy ratio of each region of interest, as well as the importance weight of each region of interest are collected from the anomaly report. Then, according to the missing area and color abnormality ratio of each region of interest, the anomaly weight is calculated for each region of interest. The anomaly weight calculation formula is:

[0060] ; represents the anomaly weight of the i-th region of interest, represents the missing area occupancy ratio of the i-th region of interest, represents the proportion of the color abnormal area in the i-th region of interest, , They are all preset non-zero proportional coefficients, which are used to adjust the influence of the missing area occupancy ratio and the color abnormal area occupancy ratio on the abnormal weight; By comprehensively considering the proportion of missing areas and color abnormal areas, the degree of abnormality of each region of interest was quantified;

[0061] Then the relative position weight is calculated for each region of interest. The relative position weight calculation formula is:

[0062] ; represents the relative position weight of the i-th region of interest, It indicates the maximum distance between the center point of all the regions of interest in the target detection area and the preset image center of the optical element. Indicates the distance between the center point of the i-th region of interest in the target detection area and the preset image center of the optical element; It reflects the importance of the location of the region of interest in the image. The closer the area is to the center, the higher the weight is, ensuring that the central area of ​​the image (usually the area of ​​most concern) has a greater impact on the overall evaluation;

[0063] Finally, the importance weight, relative position weight, and anomaly weight of each region of interest are combined to calculate the image acquisition anomaly index of the target detection area. The calculation formula is:

[0064] ; represents the importance weight of the i-th region of interest, m is the total number of regions of interest divided into the image of the target detection area, The image acquisition anomaly index of the target detection area comprehensively considers the importance weight, relative position weight and anomaly weight of each region of interest, quantifies the degree of acquisition anomaly of the overall image, and reflects the possible problems and deviations in the image acquisition process. Value: Indicates that there are many abnormal areas or significant quality problems during image acquisition. Specifically: High Anomaly Weight: Many areas of interest have large missing or color anomalies. High Relative Position Weight: Areas near the center of the image have serious anomalies, so these areas are weighted higher. Low Value: indicates that there are fewer abnormalities during image acquisition, the image quality is high, the quality of each area of ​​interest is good, and there are fewer or less obvious abnormalities and problem areas.

[0065] Abnormal analysis refers to: constructing a three-dimensional model or a multi-view stitching model of the image based on the collected image data of the target detection area, integrating multi-angle image data, generating an overall view, and then performing image loss detection and color anomaly detection. Then, the detected image loss and color anomaly areas are marked on the image of the overall view to generate an abnormality report; specifically: imaging the target detection area from different angles through high-resolution camera equipment, and collecting multi-angle image data. Purpose: Obtain comprehensive image data to ensure that the images processed later cover the entire detection area. Use image stitching algorithms or three-dimensional modeling technology to integrate multi-view image data into an overall view. Image stitching: Use feature matching algorithms (such as SIFT, SURF) to detect and match feature points in adjacent images, and then stitch multiple images into a complete image through image registration and fusion technology. Three-dimensional modeling: Use structured light scanning, stereo vision or laser scanning technology to build a three-dimensional model of the target detection area, and integrate multi-angle image data into a three-dimensional structure. Purpose: Generate a complete view of the target detection area, providing a basis for subsequent image loss detection and color anomaly detection. Existing technologies: Edge detection: Use edge detection algorithms such as Canny and Sobel to identify edges in images, detect discontinuous or broken edges, and identify possible missing areas. Texture analysis: Detect areas with discontinuous textures in images through texture feature extraction (such as gray-level co-occurrence matrix and local binary pattern), which may be missing parts of the image. Blank area detection: Identify missing parts of images by analyzing low pixel density or blank areas in images. Existing technologies: Color histogram comparison: Calculate the color histogram of an image or image area and compare it with the color histogram of a normal image to detect color distribution anomalies. Color consistency detection: Use color consistency algorithms (such as histogram matching and color distribution comparison) to check the uniformity and consistency of colors in an image area and detect areas with inconsistent colors. Machine learning methods: Use trained machine learning models (such as support vector machines and random forests) to classify color features and identify color anomaly areas. Mark the detected image missing and color anomaly areas in the overall view. Generate a detailed anomaly report, including the type and location of the anomaly for each detected area.

[0066] The logic for obtaining the detection area change anomaly coefficient is as follows: the image acquisition anomaly index and the environmental change drastic index of the target detection area are substituted into the pre-trained machine learning model, and the output result of the machine learning model is the detection area change anomaly coefficient; the machine learning model is not specifically limited here, and any model that can perform a comprehensive analysis on the image acquisition anomaly index and the environmental change drastic index of the target detection area to generate the detection area change anomaly coefficient is acceptable. In order to implement the technical solution of the present invention, the present invention provides a specific implementation method: select a suitable machine learning model, such as a linear regression model, ; is the baseline value of the regression model, and the value of the abnormal coefficient of regional change when VI and ICAI are both zero. is the coefficient of the environmental change severity index (VI), which indicates the influence of VI on the abnormal change coefficient of the detection area. A negative value means that an increase in the environmental change index will lead to a decrease in the abnormal coefficient of the detection area change; if A positive value means that an increase in the environmental change index will increase the detection area change anomaly coefficient. The coefficient of the image acquisition anomaly index (ICAI) indicates the degree of influence of ICAI on the detection area change anomaly coefficient. A negative value means that an increase in the image acquisition anomaly index will lead to a decrease in the detection area change anomaly coefficient; if A positive value means that an increase in the image acquisition anomaly index will lead to an increase in the detection area change anomaly coefficient. The detection area change anomaly coefficient BH is a comprehensive indicator that quantifies the overall change anomaly degree of the detection area by combining the influence of VI and ICAI. The larger the BH, the worse the acquisition quality of the detection area. The detection area with poor acquisition quality needs to be re-acquired to ensure that the final detection result is accurate and reliable. The detection area change anomaly coefficient is compared with the preset change anomaly threshold. If the detection area change anomaly coefficient is greater than or equal to the preset change anomaly threshold, it is divided into a non-adopted data area. If the detection area change anomaly coefficient is less than the preset change anomaly threshold, it is divided into an adopted data area.

[0067] The difference degree analysis of the environmental change drastic index of all detection areas refers to: obtaining the environmental change drastic index of all detection areas, then calculating the standard deviation of all environmental change drastic indexes, comparing the standard deviation value with the preset fluctuation threshold, if the standard deviation value is greater than or equal to the preset fluctuation threshold, both the adopted data area and the non-adopted data area need to re-collect images, if the standard deviation value is less than the preset fluctuation threshold, only the non-adopted data area is re-collected. The degree of difference in the environmental change drastic index reflects the difference in environmental changes in each detection area during the imaging process. If these differences are too large, the image quality of different areas may be inconsistent, thereby affecting the accuracy of the overall detection result. The calculated standard deviation is compared with the preset fluctuation threshold. If the standard deviation is greater than or equal to the threshold, it indicates that the difference in environmental changes between the areas is too large and it is necessary to re-collect images comprehensively; if the standard deviation is less than the threshold, only the images of the non-adopted data area can be re-collected. While ensuring data accuracy and consistency, optimize resource utilization and improve detection efficiency. Avoid unnecessary comprehensive re-collection, saving time and cost. Global re-collection: If the standard deviation of the environmental change index is large, it means that each detection area is greatly affected by environmental changes, and all detection areas need to be re-collected to ensure data consistency. Local re-collection: If the standard deviation is small, it means that the environment in most detection areas is stable, and only the non-adopted data areas need to be supplemented, which not only ensures data quality but also improves detection efficiency.

[0068] Reasoning about the optimization direction when reacquiring an image means:

[0069] Obtain the results of the environmental change drastic index, image acquisition abnormality index, and difference degree analysis of the target detection area of ​​the re-collected image respectively, define the results of the environmental change drastic index, image acquisition abnormality index, and difference degree analysis as input variables, define the output variables as the direction to be optimized, fuzzify the input variables, convert the values ​​of the input variables into fuzzy sets, fuzzify the output variables, convert the output variables into fuzzy sets, formulate fuzzy rules, describe the optimization requirements under different data type combinations, and reason the fuzzified input variables through fuzzy rules to determine the direction to be optimized. Environmental change drastic index: reflects the degree of environmental change in the detection area during the imaging process. Image acquisition abnormality index: reflects the quality abnormalities that occur during the image acquisition process. The result of the difference degree analysis: refers to the standard deviation of the environmental change drastic index of all detection areas. Convert each input variable into a fuzzy set: Environmental change drastic index: low: indicates that the environmental change is small. Medium: indicates that the environmental change is medium. High: indicates that the environmental change is large. Image acquisition abnormality index: low: indicates that the image quality is good and there is no obvious abnormality. Medium: indicates that the image quality has some problems, but not serious. High: indicates that the image quality is seriously abnormal and needs special attention. Results of the difference degree analysis: Low: Indicates that the difference in the environmental change drastic index is small, and the environmental changes in each detection area are relatively consistent. High: Indicates that the difference in the environmental change drastic index is large, and the environmental changes in each detection area are inconsistent. Fuzzy processing output variables: Convert the direction to be optimized into a fuzzy set: Environmental control optimization: Applicable to the case where the environmental change drastic index is high, by optimizing environmental control, reduce the impact of environmental changes on image acquisition. Image acquisition optimization: Applicable to the case where the image acquisition anomaly index is high, by optimizing image acquisition parameters and methods to improve image quality. Optimization of both: Applicable to the case where both the environmental change drastic index and the image acquisition anomaly index are high, perform environmental control and image acquisition optimization at the same time to ensure data quality; Describe the optimization requirements under different data type combinations:

[0070] If the environmental change drastic index is high and the image acquisition abnormality index is low, environmental control optimization is required.

[0071] If the environmental change drastic index is low and the image acquisition abnormality index is high, image acquisition optimization is required.

[0072] If the environmental change drastic index is medium and the image acquisition abnormality index is medium, image acquisition optimization is required.

[0073] If the environmental change drastic index is high and the image acquisition abnormality index is high, both need to be optimized.

[0074] If the result of the difference degree analysis is high, all detection areas need to be optimized, preferably both.

[0075] If the result of the difference degree analysis is low, corresponding optimization is performed according to the environmental change drastic index and the image acquisition abnormality index.

[0076] Finally, the optimization direction for re-collecting images is obtained. When it is not necessary to re-collect images, all collected images are summarized and used as the basis for surface contaminant detection. Rapid identification is performed according to preset contaminant identification strategies. These identification strategies are based on existing technologies and can ensure the accuracy and efficiency of the detection process. The identification strategies are not described in detail here.

[0077] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0078] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0079] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A rapid detection method for contaminants on the surface of optical components based on visual imaging, characterized in that: The following steps are involved: Dividing the surface of the optical element into a plurality of detection areas and performing imaging acquisition on each detection area; Analyze the information during the imaging acquisition process of each detection area to generate an environmental change drastic index and an image acquisition abnormality index respectively; Based on the comprehensive analysis of the environmental change intensity index and the image acquisition abnormality index, the abnormal change coefficient of the detection area is generated; Based on the abnormal coefficient of change in the detection area, each detection area is divided into one of two area types: an adopted data area or a non-adopted data area; Perform difference analysis on the environmental change severity index of all detection areas, and judge the area type for re-image acquisition based on the difference analysis results and re-acquire images; Based on the results of the environmental change intensity index, image acquisition abnormality index, and difference degree analysis, the optimization direction when re-acquiring images is inferred; When there is no need to re-collect images, all collected images are summarized and used as the basis for surface contaminant detection; The difference degree analysis of the environmental change severity index of all detection areas refers to: obtaining the environmental change severity index of all detection areas, then calculating the standard deviation of all environmental change severity indexes, and comparing the standard deviation of all environmental change severity indexes with a preset fluctuation threshold. If the standard deviation of all environmental change severity indexes is greater than or equal to the preset fluctuation threshold, both the adopted data area and the non-adopted data area need to re-collect images. If the standard deviation of all environmental change severity indexes is less than the preset fluctuation threshold, only the non-adopted data area needs to re-collect images.

2. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 1, characterized in that: The environmental change severity index is determined based on the target indicators in the environmental conditions. The determination logic of the target indicators is: Data is collected for each preset indicator in the environmental factors, and the correlation coefficient between each indicator in the environmental factors and the image quality is calculated. The calculation formula is: ; represents the value of variable x, represents the average value of variable x, which is an indicator of environmental factors. represents the value of variable y, represents the average value of variable y, variable y is the image quality, Denotes the correlation coefficient, and then calculates the correlation coefficient The absolute value of , select the indicator corresponding to the largest absolute value as the target indicator.

3. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 2, characterized in that: A preset meshing method is used to divide the surface of the optical component into multiple detection areas.

4. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 3, characterized in that: The logic for obtaining the environmental change drastic index is: Collect the target index data of the environmental conditions of each detection area during the imaging acquisition process to obtain the time series P of the target detection area: ; n is the total number of target indicator data, Indicates the target indicator data corresponding to the number n, and then calculates the mean of the time series P , the formula for calculating the environmental change intensity index is: ; Indicates the drastic index of environmental changes in the target detection area. Represents the result of the normalization of the i-th data point, satisfying: ; n is the total number of target indicator data, represents the volatility weight of the ith data point, represents the outlier weight of the ith data point, The indicator function representing the normalization result of the i-th data point, represents the local standard deviation of the 1st to the i-th data point, Represents the target indicator data of the i-th data point, represents the standard deviation of the time series P, Represents the standard deviation change rate weight of the ith data point.

5. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 4, characterized in that: A basic threshold is preset. When the absolute value of the result of the data point normalization processing is greater than or equal to the preset basic threshold, the data point is identified as an abnormal data point, and its corresponding target indicator data is an abnormal value. The output result is 1; when the absolute value of the result of the data point standardization is less than the preset basic threshold, the data point is identified as a normal data point, and its corresponding target indicator data is a normal value. The output result is 0.

6. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 5, characterized in that: The logic for obtaining the image acquisition anomaly index is: First, the image of the target detection area is divided into multiple regions of interest according to the preset image segmentation strategy, and then anomaly analysis is performed to generate an anomaly report. The missing area occupancy ratio and color abnormal area occupancy ratio of each region of interest, as well as the importance weight of each region of interest are collected from the anomaly report. Then, according to the missing area occupancy ratio and color abnormal area occupancy ratio of each region of interest, the anomaly weight is calculated for each region of interest. The anomaly weight calculation formula is: ; represents the anomaly weight of the jth region of interest, represents the missing area occupancy ratio of the jth region of interest, represents the proportion of the color abnormal area in the jth region of interest, , They are all preset non-zero proportional coefficients, which are used to adjust the influence of the missing area occupancy ratio and the color abnormal area occupancy ratio on the abnormal weight; Then the relative position weight is calculated for each region of interest. The relative position weight calculation formula is: ; represents the relative position weight of the jth region of interest, It indicates the maximum distance between the center point of all the regions of interest in the target detection area and the preset image center of the optical element. Indicates the distance between the center point of the jth region of interest in the target detection area and the preset image center of the optical element; Finally, the importance weight, relative position weight, and anomaly weight of each region of interest are combined to calculate the image acquisition anomaly index of the target detection area. The calculation formula is: ; represents the importance weight of the jth region of interest, m is the total number of regions of interest divided into the image of the target detection area, Indicates the image acquisition anomaly index of the target detection area.

7. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 6, characterized in that: Abnormal analysis means: generating an overall view based on the collected image data of the target detection area, then performing image missing detection and color abnormality detection, and then marking the detected image missing and color abnormality areas on the image of the overall view to generate an abnormality report.

8. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 7, characterized in that: The logic for obtaining the abnormal coefficient of detection area change is as follows: the image acquisition abnormality index and the environmental change drastic index of the target detection area are substituted into the pre-trained machine learning model. The output result of the machine learning model is the abnormal coefficient of detection area change. The detection area change anomaly coefficient is compared with the preset change anomaly threshold. If the detection area change anomaly coefficient is greater than or equal to the preset change anomaly threshold, it is divided into a non-adopted data area. If the detection area change anomaly coefficient is less than the preset change anomaly threshold, it is divided into an adopted data area.

9. The method for rapid detection of contaminants on the surface of optical components based on visual imaging according to claim 8, characterized in that: Reasoning about the optimization direction when reacquiring an image means: The environmental change drastic index, image acquisition anomaly index and difference degree analysis results of the target detection area of ​​the re-captured image are obtained respectively, the environmental change drastic index, image acquisition anomaly index and difference degree analysis results are defined as input variables, and the output variable is defined as the direction to be optimized. The input variable is fuzzified and the value of the input variable is converted into a fuzzy set. The output variable is fuzzified and the output variable is converted into a fuzzy set. Fuzzy rules are formulated to describe the optimization requirements under different data type combinations. The fuzzified input variables are inferred through fuzzy rules to determine the direction to be optimized.

Images