A method for online detection of particles on the surface of precision optical components
By dividing the grayscale image of the optical element surface, analyzing the electrostatic adsorption anomalies and noise impact values, and performing adaptive filtering, the problem of noise interference in the detection of particle objects on the surface of the optical element is solved and the detection accuracy is improved.
Patent Information
- Application Number
- CN202411118248.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-08-15
AI Technical Summary
The prior art cannot effectively eliminate noise interference when detecting particles on the surface of optical elements, resulting in low accuracy of detection results.
By acquiring the grayscale image of the optical element to be measured, dividing it into multiple image blocks, analyzing the electrostatic adsorption anomalies and noise influence values of each image block, adaptive filtering is performed to remove noise, and finally particles are performed in the denoising image Test.
It effectively removes noise interference, improves the accuracy of particle detection, and ensures the reliability of particle detection results on the surface of optical components.
Smart Images

Figure CN118644487B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image denoising, and in particular to an online detection method for particles on the surface of a precision optical element. Background Art
[0002] With the rapid development of science and technology, optical components are increasingly used in many fields, such as aerospace, medical diagnosis, semiconductor manufacturing, and communication technology. In these high-precision and high-demand fields, the surface quality of optical components plays a vital role in their performance. The adhesion of surface particles is one of the main factors affecting the performance of optical components. Its presence may cause problems such as changes in the light path, enhanced scattering, and energy loss, which seriously affects the stability and accuracy of the optical system. Therefore, it is necessary to detect the adhesion of particles on the surface of precision optical components to improve the quality of optical components.
[0003] The prior art usually uses image processing to detect the attachment of particles on the surface of optical components, that is, to collect images of the surface of the optical component, and then perform image recognition to obtain information related to the particles on the surface of the optical component. When determining the position of particles on the surface of the optical component, edge detection is usually performed on the surface image of the optical component to determine the position of the particles. However, due to the influence of the detection environment, image acquisition equipment, etc., noise interference will be present in the acquired surface image of the optical component, and the noise points are very similar to the particles in morphology. Therefore, directly performing edge detection on the surface image of the optical component to determine the position of the particles cannot eliminate the influence of noise, which will cause the problem of low accuracy of the detection result. Summary of the invention
[0004] In order to solve the technical problem that the position of particles can be determined by directly performing edge detection on the surface image of an optical element, the interference of noise cannot be effectively eliminated, resulting in low accuracy of the detection result, the purpose of the present invention is to provide an online detection method for particles on the surface of a precision optical element, and the technical solution adopted is as follows:
[0005] Acquire a grayscale image of the surface of the optical element to be measured when it is working, and divide the grayscale image into a plurality of image blocks;
[0006] Determine the suspected particle area according to the grayscale distribution characteristics of the pixel points in the grayscale image; analyze the position distribution characteristics of the suspected particle area in the image block, the area of the suspected particle area and the position of the image block in the grayscale image to determine the electrostatic adsorption anomaly of each image block;
[0007] Based on the aggregation of suspected particle areas in the image block, determine the noise impact value corresponding to each image block;
[0008] According to the electrostatic adsorption anomaly and noise influence value of all image blocks, the grayscale image is denoised to obtain a denoised image; and particle detection is performed on the denoised image.
[0009] Furthermore, the method for obtaining the electrostatic adsorption anomaly includes:
[0010] Select any image block as the block to be tested, determine the first electrostatic adsorption factor of the block to be tested based on the area ratio of the suspected particle area in the block to be tested, take the distance between the centroid of the block to be tested and the centroid of the grayscale image as the second electrostatic adsorption factor, and obtain the electrostatic adsorption strength of the block to be tested by combining the first electrostatic adsorption factor and the second electrostatic adsorption factor, and the first electrostatic adsorption factor is positively correlated with the electrostatic adsorption strength, and the second electrostatic adsorption factor is negatively correlated with the electrostatic adsorption strength;
[0011] Determine an ideal electrostatic adsorption characteristic value of the block to be tested based on the electrostatic adsorption strength of the block and the relative position relationship between the block to be tested and other image blocks;
[0012] Determine the actual electrostatic adsorption characteristic value of the block to be tested based on the degree of position distribution disorder of the suspected particle area in the block to be tested and the area of each suspected particle area;
[0013] The difference between the rational electrostatic adsorption characteristic value and the actual electrostatic adsorption characteristic value of the block to be tested is analyzed to obtain the electrostatic adsorption anomaly of the block to be tested.
[0014] Furthermore, the method for obtaining the ideal electrostatic adsorption characteristic value includes:
[0015] The image blocks other than the block to be tested are used as interference blocks;
[0016] The distance between the centroid of the block to be tested and the centroid of each interference block is used as a distance factor; the position weight of each interference block is obtained according to the electrostatic adsorption strength of each interference block and the distance factor corresponding to each interference block, and the electrostatic adsorption strength is positively correlated with the position weight, and the distance factor is negatively correlated with the position weight;
[0017] Obtain the straight line connecting the center of mass of the block to be tested and the center of mass of each interference block. Taking the horizontal line as the reference, the angle formed by the straight line and the horizontal line in the clockwise direction is used as the relative position eigenvalue of each interference block. Use the position weights of all interference blocks to perform weighted fusion on the corresponding relative position eigenvalues to obtain the ideal electrostatic adsorption eigenvalue corresponding to the block to be tested.
[0018] Furthermore, the method for obtaining the actual electrostatic adsorption characteristic value includes:
[0019] In the block to be tested, a straight line connecting the centroid of the block to be tested and the centroid of each suspected particle matter region is obtained, and the angle formed by the straight line and the horizontal line in the clockwise direction is used as the relative position distribution value of each suspected particle matter region with the horizontal line as a reference; the distance between the centroid of each suspected particle matter region and the centroid of the block to be tested is used as the distance parameter of each suspected particle matter region;
[0020] The product of the relative position distribution value, distance parameter and area of each suspected particle area is taken as the electrostatic adsorption influence value of each suspected particle area;
[0021] The actual electrostatic adsorption characteristic value of the block to be tested is obtained by integrating the electrostatic adsorption influence values of all suspected particle areas in the block to be tested.
[0022] Furthermore, the method for obtaining the noise impact value includes:
[0023] In any image block, the suspected particle area with the largest area in the image block is used as the reference area, and the remaining suspected particle areas are used as the areas to be analyzed;
[0024] Determine the clustering factor of the image block according to the difference between the areas of all the regions to be analyzed and the reference area;
[0025] Determine the noise impact factor of the image block based on the relative position relationship between the region to be analyzed and the reference region and the area of the region to be analyzed;
[0026] According to the clustering factor and the noise impact factor, a noise impact value of the image block is obtained, and the clustering factor and the noise impact factor both change in the same direction as the noise impact value.
[0027] Furthermore, the method for obtaining the clustering factor includes:
[0028] The difference between the area of each area to be analyzed and the area of the reference area is used as the area difference factor of each area to be analyzed;
[0029] The area difference factors of all the regions to be analyzed are averaged to obtain the clustering factor.
[0030] Furthermore, the method for obtaining the noise impact factor includes:
[0031] Obtaining the distance between the centroid of each area to be analyzed and the centroid of the reference area as a distance index of each area to be analyzed; determining an abnormality parameter of each area to be analyzed based on the distance index and area of each area to be analyzed, wherein both the distance index and the area change inversely with the abnormality parameter;
[0032] The abnormal parameters of all the areas to be analyzed are averaged to obtain the noise impact factor of the image block.
[0033] Furthermore, the method for obtaining the denoised image includes:
[0034] The electrostatic adsorption anomaly and the noise influence value of each image block are integrated to obtain an adaptive filter factor of each image block, and the electrostatic adsorption anomaly and the noise influence value are both positively correlated with the adaptive filter factor;
[0035] The adaptive filtering factor of each image block is used as the scale factor of the Wiener filter, and an adaptive filtering operation is performed on each image block to obtain a denoised image block of each image block;
[0036] A denoised image corresponding to the grayscale image is obtained based on the denoised image block.
[0037] Further, the performing particle detection on the denoised image includes:
[0038] Perform Canny edge detection on the denoised image to obtain the particle area.
[0039] Furthermore, the method for obtaining the suspected particle matter area includes:
[0040] A grayscale image of the surface of the optical element without particles attached is used as a reference image;
[0041] The grayscale image of the surface of the optical element to be tested is compared with the reference image, and the pixels whose grayscale value difference is greater than a preset grayscale threshold are regarded as suspected particle pixels. A connected domain analysis is performed on the suspected particle pixels in the grayscale image of the surface of the optical element to be tested to obtain the suspected particle area.
[0042] The present invention has the following beneficial effects:
[0043] First, a grayscale image of the surface of the optical element to be tested is obtained, and the grayscale image is divided into regions to obtain multiple image blocks. In the subsequent process, the denoising strength can be analyzed for different image blocks. Since the existence of noise points and particles will cause the grayscale value of the pixel point to change to a certain extent, the suspected particle area can be determined based on the grayscale distribution characteristics of the pixel points in the grayscale image. Because electrostatic adsorption will affect the position of the particles, and the electrostatic adsorption capabilities of different image blocks are also different, while the distribution of noise points is not affected by electrostatic adsorption, the electrostatic adsorption anomaly of each image block can be determined based on the position distribution characteristic value, area size and position of the suspected particle area in the image block. The electrostatic adsorption anomaly can be used as an indicator to characterize the degree of noise influence in the image block; further, since electrostatic adsorption will cause the position of the particles to be relatively concentrated, while the distribution of noise points is more random and discrete, the noise influence value of the image block is determined based on the distribution and aggregation of the suspected particle area in the image block, and the noise influence value can also be used as an indicator to characterize the degree of noise influence in the image block. Finally, the electrostatic adsorption anomaly and the noise influence value of the image block are combined to determine the noise influence degree of each image block and denoise it. The denoised image corresponding to the grayscale image can be obtained. The denoised image at this time can effectively remove noise interference, so the detection of particles in the denoised image can obtain more accurate detection results. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0045] Figure 1 A method flow chart of an online detection method for particles on the surface of a precision optical element provided by an embodiment of the present invention;
[0046] Figure 2 A method flow chart of a method for obtaining an anomaly of electrostatic adsorption provided by an embodiment of the present invention;
[0047] Figure 3 A method flow chart of a method for obtaining a noise impact value provided by an embodiment of the present invention;
[0048] Figure 4 A system block diagram of an online detection system for particles on the surface of a precision optical element provided by an embodiment of the present invention;
[0049] Figure 5 A schematic diagram of the structure of a computer device provided by an embodiment of the present invention;
[0050] Figure 6 A schematic diagram of a computer-readable storage medium provided for one embodiment of the present invention. DETAILED DESCRIPTION
[0051] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of an online detection method for particles on the surface of a precision optical element proposed by the present invention, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0052] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0053] The specific scheme of the on-line detection method for particles on the surface of a precision optical element provided by the present invention is described in detail below with reference to the accompanying drawings.
[0054] See also Figure 1 , which shows a method flow chart of an online detection method for particles on the surface of a precision optical element provided by an embodiment of the present invention, the method comprising the following steps:
[0055] Step S1: obtaining a grayscale image of the surface of the optical element to be measured when it is working, and dividing the grayscale image into a plurality of image blocks.
[0056] During the use of precision optical components, due to the influence of tiny particles from the environment and equipment, particles will be attached to the surface of the optical components. The presence of particles may cause problems such as changes in the light path and enhanced scattering, thereby affecting the performance of the optical components. Therefore, it is necessary to detect particles on the surface of precision optical components to ensure the stability of the working state of the optical components.
[0057] First, it is necessary to obtain an image of the surface of the optical element to be tested, and then analyze and process the image. In this embodiment of the present invention, a high-resolution image acquisition device can be used to photograph the surface of the precision optical element during use. The specific image acquisition device can use an industrial CCD camera, and the brightness of the surrounding environment must be ensured during shooting. At this point, an RGB image of the surface of the optical element to be tested can be obtained.
[0058] In the process of particle detection, since the main concern is the location, size and distribution of the particles, and there is no need to perform color analysis on the particles, we can focus on the processing and analysis of grayscale images. Grayscale images not only simplify the image processing process, but also reduce the computational complexity, and can more clearly highlight the key features. Therefore, the RGB image of the surface of the optical element to be tested is grayscaled to obtain a grayscale image of the surface of the optical element to be tested. It should be noted that when graying the RGB image, the maximum value method, weighted grayscale or average grayscale method can be used, and they are all technical means well known to those skilled in the art, and are not limited or elaborated here; at the same time, the image acquisition equipment and devices can also be adjusted according to the implementation scenario, which is not limited here.
[0059] There are multiple functional areas inside the optical element. Each functional area is designed according to its specific purpose and undertakes different tasks. These functional areas will generate current when working, and due to factors such as static charges in the environment and the charge characteristics of the material itself, the functional areas will have the ability to adsorb surrounding tiny particles. The electrostatic adsorption capacity of different functional areas will also be different. Therefore, the grayscale image is divided according to the functional area of the optical element to obtain multiple image blocks. In the subsequent process, each image block can be analyzed separately to determine the noise impact on each individual image block and the distribution of particles.
[0060] It should be noted that in this embodiment of the present invention, the optical element specifically takes an integrated optical chip as an example; and in other embodiments of the present invention, if the small components inside the optical element are arranged regularly, the grayscale image can also be divided into 9 image blocks, and the specific values are not limited.
[0061] Step S2: Determine the suspected particle area according to the grayscale distribution characteristics of the pixel points in the grayscale image; analyze the position distribution characteristics of the suspected particle area in the image block, the area of the suspected particle area and the position of the image block in the grayscale image to determine the electrostatic adsorption anomaly of each image block.
[0062] In the embodiment of the invention, the real particle area and noise point on the surface of the optical element are collectively referred to as the suspected particle area. Due to the presence of particles and noise on the surface of the optical element, the original grayscale characteristics will mutate, so the suspected particle area can be screened out based on the grayscale distribution characteristics. Further, because the current of different functional areas inside the optical element, such as the integrated optical chip in this embodiment of the present invention, is inconsistent during operation, resulting in uneven charge distribution, which will generate an electrostatic field, thereby having an adsorption effect on the particles, but this effect has no effect on the distribution of noise. At the same time, the possibility of functional areas at different positions contacting particles is also different. Therefore, the electrostatic adsorption anomaly of each image block can be determined based on the position distribution characteristics, area size and position distribution of the suspected particle area in the image block in the grayscale image. The electrostatic adsorption anomaly can be used to characterize the difference between the actual electrostatic adsorption capacity and the ideal electrostatic adsorption capacity of each image block, that is, the difference between the position distribution characteristics of the suspected particle area in actual conditions and the distribution characteristics that the particle area should have in ideal conditions, so as to evaluate the degree of noise interference of the image block.
[0063] First, it is necessary to obtain the suspected particle area in the grayscale image. Since both the particle area and the noise point will cause the grayscale value of the pixel to change suddenly, the suspected particle area is screened out in the grayscale image based on the grayscale features.
[0064] Preferably, in one embodiment of the present invention, the method for obtaining the suspected particle matter area includes:
[0065] The grayscale image of the surface of the optical element without particles attached can be used as a reference image; then the grayscale image of the surface of the optical element to be tested is compared with the reference image. Since the presence of particles and noise points will cause a sudden change in the grayscale value, the pixels whose grayscale value difference is greater than the preset grayscale threshold are taken as suspected particle pixels. Finally, the suspected particle pixels in the grayscale image of the surface of the optical element to be tested are subjected to a connected domain analysis to obtain the suspected particle area. It should be noted that the preset grayscale threshold is set to 5, and the specific value can be adjusted according to the implementation scenario, which is not limited here; the connected domain analysis process is a technical process well known to those skilled in the art and will not be described in detail here.
[0066] In other embodiments of the present invention, the suspected particle area may be obtained by the following method.
[0067] (1) Based on edge detection algorithms such as Canny, Sobel, and Prewitt, it is possible to detect areas with obvious grayscale changes in the grayscale image of the surface of the optical element to be tested, and these areas may be areas where particles are attached or where noise exists, so the areas detected by edge detection can be used as suspected particle areas. It should be noted that the edge detection algorithm is not limited here, and the specific implementation process of each edge detection algorithm is a process well known to those skilled in the art, and will not be described in detail here.
[0068] (2) A pre-trained neural network, such as a CNN neural network, can also be used to identify the grayscale image of the surface of the optical element to be tested, thereby obtaining the suspected particle area. It should be noted that the training method of the neural network is a process well known to those skilled in the art and will not be described in detail here.
[0069] During the operation of the optical element, an electric current will be generated inside. At this time, the current will cause uneven charge distribution, and then generate an electrostatic field, causing the surrounding tiny particles to adhere to the surface of the optical element. The presence of these tiny particles will affect the transmission and processing of optical signals, and even cause the performance of the optical element to deteriorate. The size and distribution of the internal current in different functional areas of the optical element will also be different, resulting in different electrostatic field strengths and different adsorption capabilities for particles. For example, some functional blocks may generate larger currents and stronger electrostatic fields due to the need to process stronger optical signals or perform more complex tasks, so they may have stronger adsorption capabilities for particles. However, the electrostatic adsorption capacity will not have an adsorption effect on noise. Therefore, in the process of analyzing the electrostatic adsorption capacity of the image block corresponding to each functional area, the electrostatic adsorption anomaly of each image block can be determined mainly based on the position distribution characteristics of the suspected particle area in the image block, the area of the suspected particle area, and the position of the image block in the grayscale image. The electrostatic adsorption anomaly is actually a quantitative indicator that can indirectly evaluate the degree to which the image block is disturbed by noise.
[0070] Preferably, in one embodiment of the present invention, the method for obtaining the electrostatic adsorption anomaly comprises:
[0071] See also Figure 2 , which shows a method flow chart of a method for obtaining an anomaly of electrostatic adsorption in one embodiment of the present invention, the method comprising the following steps:
[0072] Step S201: determining the electrostatic adsorption strength of each image block based on the area of the suspected particle region in each image block and the position in the grayscale image.
[0073] The electrostatic adsorption strength of different functional areas on particles is mainly reflected in the area size of the suspected particle area. The larger the area of the suspected particle area, the stronger the adsorption capacity. At the same time, when particles contact the optical element, due to the influence of their movement trajectory or diffusion mode, they may preferentially contact the functional area close to the outside of the optical element. This contact order means that the functional area close to the outside has the opportunity to contact the particles first. Therefore, there will be deviations in quantifying the electrostatic adsorption strength only by using the area size of the suspected particle area, and the influence of the position of the functional area needs to be eliminated.
[0074] For any image block, a first electrostatic adsorption factor of the image block is determined based on the area ratio of the suspected particle area in the image block.
[0075] Whether the functional area is close to the outside of the optical element can be measured by the distance between the centroid of the image block and the centroid of the grayscale image, so the distance between the centroid of the image block and the centroid of the grayscale image is used as the second electrostatic adsorption factor.
[0076] Finally, the first electrostatic adsorption factor and the second electrostatic adsorption factor are combined to obtain the electrostatic adsorption strength of the image block. The formula model of the electrostatic adsorption strength includes:
[0077] ;
[0078] in, Indicates The electrostatic adsorption strength of each image block; Indicates In the image block The area of suspected particle matter areas; Indicates The number of suspected particle areas in an image block; Indicates The area of the image block; Indicates A second electrostatic adsorption factor corresponding to an image block; Indicates the preset first parameter.
[0079] In the formula model of electrostatic adsorption strength, the ratio of the total area of all suspected particle areas in each image block to the image block area is calculated as the first electrostatic adsorption factor , when the value is larger, it means that there are more suspected particles in the image block, which indicates that the electrostatic adsorption ability of the image block is stronger to a certain extent; at the same time, in order to more accurately quantify the electrostatic adsorption ability of each image block, the interference of the position factor of the image block is eliminated, and the position of the image block is quantified by the distance between the centroid of the image block and the centroid of the grayscale image to obtain the second electrostatic adsorption factor. The larger the second electrostatic adsorption factor is, the closer the image block is to the edge of the grayscale image, and the higher the priority of the image block in contacting the particles, so the confidence of the aforementioned first electrostatic adsorption factor will be reduced. On the contrary, if the second electrostatic adsorption factor is smaller, it means that the image block is farther away from the edge of the grayscale image, and the confidence of the aforementioned first electrostatic adsorption factor is higher. Based on the above logic, the value after the second electrostatic adsorption factor is negatively correlated and mapped is multiplied by the first electrostatic adsorption factor to obtain the electrostatic adsorption strength of the image block.
[0080] It should be noted that the first parameter is preset The role of is to prevent the denominator from being 0, and the value here can be 0.001. The specific value can be adjusted according to the implementation scenario and is not limited here. The negative correlation relationship means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtraction relationship, a division relationship, etc., so in other embodiments of the present invention, when performing negative correlation mapping on the second electrostatic adsorption factor, other methods can also be used, which are not limited here.
[0081] Step S202: randomly select an image block as a block to be tested, and determine an ideal electrostatic adsorption characteristic value of the block to be tested based on the electrostatic adsorption strength of the image block and the relative position relationship between the block to be tested and other image blocks.
[0082] When the influence of noise is not analyzed, because different functional areas have different adsorption capabilities for particles, during the use of the optical element, the distribution of particles in a certain functional area will be affected by the combined effect of the adsorption capabilities of multiple functional areas, resulting in a distribution feature. The distribution feature can be quantified by the position distribution of the functional areas. In an embodiment of the present invention, the distribution feature is used to characterize the ideal electrostatic adsorption situation of the image block to obtain the ideal electrostatic adsorption characteristic value.
[0083] The image blocks other than the block to be tested are regarded as interference blocks.
[0084] Since the interference blocks that are closer to the block to be tested may have a greater adsorption effect on the particles on the block to be tested, the distance between the center of mass of the block to be tested and the center of mass of each interference block is taken as the distance factor. At the same time, the electrostatic adsorption intensity of each interference block and the distance factor corresponding to each interference block are combined to obtain the position weight of each interference block. The electrostatic adsorption intensity is positively correlated with the position weight, and the distance factor is negatively correlated with the position weight.
[0085] Then analyze the relative position between the interference block and the block to be tested, obtain the straight line where the centroid of the block to be tested and the centroid of each interference block are located, and take the angle formed by the straight line and the horizontal line in the clockwise direction as the reference. , as the relative position feature value of each interference block; then use the position weights of all interference blocks to perform weighted fusion on the corresponding relative position feature values to obtain the ideal electrostatic adsorption feature value corresponding to the block to be tested. For example, the formula model of the ideal electrostatic adsorption characteristic value includes:
[0086] ;
[0087] in, Indicates the block to be tested The ideal electrostatic adsorption characteristic value of Represents the total number of image blocks in the grayscale image; Indicates the block to be tested No. The relative position feature value corresponding to the interference block; Indicates the block to be tested No. The electrostatic adsorption strength of each interference block; Indicates the block to be tested No. The distance factor corresponding to the interference block; represents the hyperbolic tangent function.
[0088] In the formula model of the ideal electrostatic adsorption characteristic value, the angle formed by the straight line connecting the center of mass of the block to be tested and the center of mass of the interference block and the horizontal line in the clockwise direction is calculated, and the value range is , and obtain the relative position characteristic value, which can characterize the position distribution of the interference block of the block to be tested; when the distance between the interference block and the block to be tested is closer, the adsorption ability will be stronger, so the distance between the center of mass of the interference block and the center of mass of the block to be tested is used as the distance factor, and the distance factor is combined with the electrostatic adsorption strength of the interference block to obtain the position weight of the interference block , the larger the position weight, the greater the influence of the electrostatic adsorption ability of the interference block on the presence of particles in the block to be tested; finally, the relative position eigenvalue is weighted by the position weight of each interference block to obtain the , and finally all the interference blocks A comprehensive analysis is performed to obtain the ideal electrostatic adsorption characteristic value of the block to be tested. Based on the position distribution of the interference block corresponding to the block to be tested, the ideal electrostatic adsorption characteristic value is quantified to characterize the position distribution of particles under the action of electrostatic adsorption under ideal conditions.
[0089] In other embodiments of the present invention, the formula model of the ideal electrostatic adsorption characteristic value can also be:
[0090] ;
[0091] in, Indicates the block to be tested The ideal electrostatic adsorption characteristic value of Represents the total number of image blocks in the grayscale image; Indicates the block to be tested No. The relative position feature value corresponding to the interference block; Indicates the block to be tested No. The electrostatic adsorption strength of each interference block; Indicates the block to be tested No. The distance factor corresponding to the interference block; Represents the normalization function.
[0092] In the above formula model, when each interference block is obtained Afterwards, the values of all interference blocks can be averaged and used The function is normalized to obtain the ideal electrostatic adsorption characteristic value of the block to be tested.
[0093] It should be noted that the normalization method may also use maximum and minimum value normalization, and the specific method is not limited.
[0094] So far, by analyzing the adsorption characteristics and relative position relationship of each functional area, the ideal electrostatic adsorption characteristic value of each image block is obtained.
[0095] Step S203: determining the actual electrostatic adsorption characteristic value of the block to be tested based on the degree of position distribution disorder of the suspected particle area in the block to be tested and the area of each suspected particle area.
[0096] During the actual operation of the optical element, under the influence of noise, the suspected particle area in each block to be tested will have a distribution feature, and the distribution feature can be quantified by the position distribution of the suspected particle area in the block to be tested. In an embodiment of the present invention, the distribution feature is used to characterize the actual electrostatic adsorption situation of the image block to obtain the actual electrostatic adsorption characteristic value.
[0097] First, in the block to be tested, obtain the straight line connecting the center of mass of the block to be tested and the center of mass of each suspected particle matter area. Taking the horizontal line as the reference, the angle formed by the straight line and the horizontal line in the clockwise direction is used as the relative position distribution value of each suspected particle matter area; the distance between the center of mass of each suspected particle matter area and the center of mass of the block to be tested is used as the distance parameter of each suspected particle matter area.
[0098] Then, the product of the relative position distribution value, the distance parameter and the area of each suspected particle area is used as the electrostatic adsorption influence value of each suspected particle area.
[0099] Finally, the electrostatic adsorption influence values of all suspected particle areas in the block to be tested are combined to obtain the actual electrostatic adsorption characteristic value of the block to be tested. For example, the formula model of the actual electrostatic adsorption characteristic value includes:
[0100] ;
[0101] in, Indicates the block to be tested The actual electrostatic adsorption characteristic value; Indicates the block to be tested Middle The area of suspected particle matter areas; Indicates the block to be tested The number of suspected particle areas in the Indicates the block to be tested No. The relative position distribution value corresponding to the suspected particle area; Indicates the block to be tested No. The distance parameter corresponding to the suspected particle area; represents the hyperbolic tangent function.
[0102] In the formula model of the actual electrostatic adsorption characteristic value, the relative position distribution value is obtained by calculating the angle formed by the straight line connecting the centroid of the block to be tested and the centroid of each suspected particle matter area and the horizontal line in the clockwise direction. This value can characterize the position distribution of the suspected particle matter area in the block to be tested; similarly, the distance between the suspected particle matter area and the block to be tested can also characterize the position distribution of the suspected particle matter area, so the distance between the centroid of each suspected particle matter area and the centroid of the block to be tested is calculated as the distance parameter; and the area size of each suspected particle matter area can characterize the electrostatic adsorption capacity in the block to be tested, so finally the product of the relative position distribution value, distance parameter and area of each suspected particle matter area is taken as the electrostatic adsorption influence value of each suspected particle matter area. Finally, the electrostatic adsorption influence values of all suspected particle areas in the block to be tested are integrated to obtain the actual electrostatic adsorption characteristic value of the block to be tested; based on the position distribution between the suspected particle areas in the block to be tested, the actual electrostatic adsorption characteristic value is quantified to characterize the position distribution of the suspected particle areas under the action of electrostatic adsorption in actual situations.
[0103] In other embodiments of the present invention, the average of the electrostatic adsorption influence values of all suspected particle areas in the block to be tested can also be normalized as the actual electrostatic adsorption characteristic value of the block to be tested; the normalization process can specifically use maximum and minimum value normalization and Normalization functions, etc., are not limited here.
[0104] Step S204: determining the electrostatic adsorption anomaly of the block to be tested based on the ideal electrostatic adsorption characteristic value and the actual electrostatic adsorption characteristic value of the block to be tested.
[0105] Due to the influence of noise, there will be a deviation between the position distribution of the suspected particle area in the block to be tested and the position distribution of the surrounding interference blocks. Since noise will not be affected by the adsorption effect, the deviation can to some extent characterize the degree of influence of noise on the block to be tested. In the embodiment of the present invention, the deviation is used as the electrostatic adsorption anomaly. For example, the formula model of electrostatic adsorption anomaly includes:
[0106] ;
[0107] in, Indicates the block to be tested The electrostatic adsorption anomaly; Indicates the block to be tested The actual electrostatic adsorption characteristic value; Indicates the block to be tested The ideal electrostatic adsorption characteristic value.
[0108] In the formula model of electrostatic adsorption anomaly, the difference between the rational electrostatic adsorption eigenvalue and the actual electrostatic adsorption eigenvalue of the block to be tested is analyzed to obtain the electrostatic adsorption anomaly of the block to be tested. The larger the value, the more noise there is in the block to be tested, that is, the greater the degree of noise influence.
[0109] So far, the electrostatic adsorption anomaly of each image block can be obtained through the above process, which is used as an indicator to characterize the influence of noise on each image block.
[0110] Step S3: Based on the aggregation of suspected particle areas in the image block, determine the noise impact value corresponding to each image block.
[0111] In the above steps, the interaction of the electrostatic adsorption capabilities between different functional areas was analyzed, and an indicator describing the influence of noise on the image block, the electrostatic adsorption anomaly, was obtained. However, since the interference of noise cannot be completely avoided in the process of calculating the electrostatic adsorption anomaly, further analysis is needed to improve the accuracy of the final result.
[0112] In view of the electrostatic adsorption ability, the particles in the functional area will form a larger area of aggregation, and only some distant unclustered particles will exist. However, the noise points will not be affected by the electrostatic adsorption, and thus have a more discrete and random distribution feature. Therefore, based on this feature, the aggregation of the suspected particle area can be analyzed in each image block, and the noise impact value corresponding to each image block can be determined as another indicator for judging whether the image block is affected by noise.
[0113] Preferably, in one embodiment of the present invention, the method for obtaining the noise impact value includes:
[0114] See also Figure 3 , which shows a method flow chart of a method for obtaining a noise impact value in one embodiment of the present invention, the method comprising the following steps:
[0115] Step S301: in each image block, determine a reference area and an area to be analyzed.
[0116] In each image block, the suspected particle area with the largest area is the most representative and has the highest probability of being the real particle aggregation area. Therefore, in any image block, the suspected particle area with the largest area in the image block is used as the reference area, and the remaining suspected particle areas are used as the areas to be analyzed.
[0117] Step S302: In each image block, a clustering factor of each image block is determined according to the difference between the areas of all the regions to be analyzed and the reference regions.
[0118] The difference between the area of each area to be analyzed and the area of the reference area is taken as the area difference factor of each area to be analyzed; the area difference factor can directly reflect the difference in area size between each area to be analyzed and the reference area, thereby indirectly reflecting the aggregation situation of the suspected particulate matter area.
[0119] The area difference factors of all the areas to be analyzed are averaged to obtain the clustering factor. The formula model of the clustering factor can be specifically, for example:
[0120] ;
[0121] in, Indicates The clustering factor of the image blocks; Indicates The number of suspected particle areas in an image block; Indicates In the image block The area of the region to be analyzed; Indicates The area of the largest suspected particle region in the image block, that is, the area of the reference region; Represents the normalization function.
[0122] In the formula model of the aggregation factor, the difference between the area of each area to be analyzed and the area of the reference area is calculated to obtain the area difference factor corresponding to each area to be analyzed. When the value is smaller, it means that the area difference between the area to be analyzed and the reference area is not large, then it is considered that the aggregation of particles in this image block is poor. On the contrary, when the value is larger, it means that the area difference between the area to be analyzed and the reference area is large, then it is considered that the aggregation of particles in this image block is good. Finally, the area difference factors of all the areas to be analyzed are averaged to obtain the aggregation factor of each image block. The larger the aggregation factor, the better the aggregation of particles in this image block.
[0123] In other embodiments of the present invention, when calculating the clustering factor of the image block, the following method may also be used:
[0124] The ratio of the area of the reference area to the area of each area to be analyzed is used as the area ratio factor of each area to be analyzed, and then the mean of the area ratio factors of all areas to be analyzed is negatively correlated and normalized to be used as the clustering factor of the image block. The formula model of the clustering factor is:
[0125] ;
[0126] in, Indicates The clustering factor of the image blocks; Indicates The number of suspected particle areas in an image block; Indicates In the image block The area of the region to be analyzed; Indicates The area of the largest suspected particle region in the image block, that is, the area of the reference region; Expressed as a natural constant An exponential function with base .
[0127] In the above formula model, the ratio of the area of the reference area to the area of each area to be analyzed is used as the area ratio factor of each area to be analyzed. , the smaller the value is, the smaller the area of other suspected particle areas that are not aggregated with the largest suspected particle area in the image block is, and it is considered that the aggregation of particles in this image block is good. On the contrary, when the value is larger, it means that the area of the area to be analyzed is not much different from that of the reference area, indicating that the aggregation of particles is poor. Finally, the area ratio factors of all the areas to be analyzed are averaged, and all the means are negatively correlated and normalized to achieve logical relationship correction, and the aggregation factor of each image block is obtained. The larger the aggregation factor is, the better the aggregation of particles in this image block is.
[0128] Step S303: In each image block, based on the relative position relationship between the region to be analyzed and the reference region and the area of the region to be analyzed, determine the noise impact factor of the image block.
[0129] Since particles are affected by electrostatic adsorption, they will aggregate, but noise points will not aggregate due to electrostatic adsorption, so noise points are more likely to exist in the area to be analyzed; and since the reference area can better represent the actual aggregation of particles, the noise impact factor of each image block can be determined by analyzing the positional relationship between the area to be analyzed and the reference area and the area size of the area to be analyzed.
[0130] The distance between the centroid of each area to be analyzed and the centroid of the reference area is obtained as the distance index of each area to be analyzed; and based on the distance index and area of each area to be analyzed, the abnormal parameters of each area to be analyzed are determined, and the distance index and area both change inversely with the abnormal parameters.
[0131] Finally, the abnormal parameters of all the areas to be analyzed are combined to obtain the noise impact factor of the image block. The formula model of the noise impact factor can be specifically, for example:
[0132] ;
[0133] in, Indicates The noise impact factor of each image block; Indicates The number of suspected particle areas in an image block; Indicates In the image block The area of the region to be analyzed; Indicates In the image block The distance index corresponding to the area to be analyzed; Indicates the preset second parameter.
[0134] In the formula model of the noise influencing factor, because the particles should gather into one area under the influence of electrostatic adsorption, the distance between the centroid of each area to be analyzed and the centroid of the reference area is calculated as the distance index. The closer the distance between a certain area to be analyzed and the reference area, that is, the smaller the distance index corresponding to the area to be analyzed, the more likely it is that it is a noise point that is not affected by electrostatic adsorption; and the smaller the area of a certain area to be analyzed, the more likely it is that it is an isolated noise point, so the distance index and the area are both placed in the denominator to obtain the abnormal parameters of each area to be analyzed. , and the larger the abnormal parameter, the higher the possibility of noise. Finally, the abnormal parameters of all the areas to be analyzed in the image block are averaged to obtain the noise impact factor of the image block.
[0135] It should be noted that the second parameter is preset The function of is to prevent the denominator from being 0. Here, the value can be 0.001. The specific value can be adjusted according to the implementation scenario and is not limited here.
[0136] Step S304: Determine the noise impact value of each image block based on the clustering factor and the noise impact factor of each image block.
[0137] Based on the analysis in the previous steps, it can be known that the larger the aggregation factor of the image block, the better the aggregation of particles in the image block. If the noise impact factor of the image block is larger under the premise that the particle aggregation in the image block is better, it can be proved that the accuracy of the noise interference in the calculated image block will be higher. Therefore, the product of the aggregation factor and the noise impact factor is normalized to obtain the noise impact value of the image block. The formula model of the noise impact value can be specifically, for example:
[0138] ;
[0139] in, Indicates The noise impact value of an image block; Indicates The clustering factor of the image blocks; Indicates The noise impact factor of each image block; represents the hyperbolic tangent function.
[0140] In the formula model of the noise impact value, the noise impact factor and the clustering factor of the image block are combined, and the product of the two is normalized to obtain the noise impact value of each image block. The larger the noise impact value, the higher the degree of noise impact on the image block.
[0141] It should be noted that since changes in the same direction represent a positive correlation, and a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases, the specific relationship can be a multiplication relationship, an addition relationship, or the power of an exponential function. Therefore, in other embodiments of the present invention, the sum of the aggregation factor and the noise impact factor can also be normalized to obtain the noise impact value.
[0142] At this point, by analyzing the aggregation of suspected particle areas in the image block, the noise impact value of each image block can be obtained, which is used as another indicator for evaluating whether the image block is affected by noise.
[0143] Step S4: De-noising the grayscale image according to the electrostatic adsorption anomaly and noise influence value of all image blocks to obtain a de-noised image; and performing particle detection on the de-noised image.
[0144] Based on the above steps, we can get two indicators to evaluate the degree to which each image block is affected by noise. Here, we combine the electrostatic adsorption anomaly and the noise impact value to denoise each image block, and get the denoised image corresponding to the grayscale image. Finally, we can detect particulate matter based on the denoised image, which can avoid noise interference to a great extent and improve detection accuracy.
[0145] Preferably, in one embodiment of the present invention, the method for obtaining a denoised image includes:
[0146] The electrostatic adsorption anomaly and noise impact value of each image block are integrated to obtain the adaptive filter factor of each image block, and both the electrostatic adsorption anomaly and the noise impact value are positively correlated with the adaptive filter factor. The formula model of the adaptive filter factor can be specifically, for example:
[0147] ;
[0148] in, Indicates Adaptive filter factor for image blocks; Indicates The noise impact value of an image block; Indicates The electrostatic adsorption anomaly of an image block.
[0149] In the formula model of the adaptive filter factor, the larger the noise impact value, the higher the noise impact on the image block; the larger the electrostatic adsorption anomaly, the more noise there is in the image block, that is, the greater the noise impact. Therefore, the product of the two is used as the adaptive filter factor of the image block.
[0150] Then, the adaptive filtering factor of each image block is used as the scale factor of the Wiener filter, and an adaptive filtering operation is performed on each image block to obtain a denoised image block of each image block; and a denoised image corresponding to the grayscale image is obtained based on the denoised image block.
[0151] It should be noted that the specific process of Wiener filtering is a technical means well known to those skilled in the art and will not be described in detail here. At the same time, since the numerical range of the proportional factor of the Wiener filter is too large or too small, it will affect the performance or stability of the filter. Therefore, in other embodiments of the present invention, after obtaining the adaptive filtering factor, its value range can also be adjusted to adapt to the needs of the filter. The specific numerical adjustment process will not be described in detail here.
[0152] In the above process, each image block in the grayscale image is adaptively filtered, so the noise differences in each image block can be eliminated in a targeted manner, and the particle density in each image block is retained; the denoised image corresponding to the grayscale image is obtained. At this time, the particle detection of the denoised image can obtain the accurate location of the particle.
[0153] Preferably, in one embodiment of the present invention, performing particle detection on the denoised image includes:
[0154] The Canny edge detection algorithm can accurately identify the edge information in the image, and most of the noise in the denoised image has been filtered out. Therefore, when Canny edge detection is applied to the denoised image, particles can be better detected. Therefore, Canny edge detection is performed on the denoised image to obtain the particle area.
[0155] In summary, firstly, a grayscale image of the surface of the optical element to be tested is obtained. Since different functional areas of the optical element have different electrostatic adsorption capabilities for particles, the grayscale image is divided according to the functional areas of the optical element to be tested to obtain multiple image blocks. In the subsequent process, the denoising strength of different image blocks can be analyzed. Since the presence of noise points and particles will cause the grayscale value of the pixel to change to a certain extent, the suspected particle area can be determined based on the grayscale distribution characteristics of the pixel points in the grayscale image. Since electrostatic adsorption will affect the position of particles, and the electrostatic adsorption capabilities of different functional blocks are also different, while the distribution of noise points is not affected by electrostatic adsorption, the electrostatic adsorption anomaly of each image block can be determined based on the position distribution characteristic value, area size and position of the suspected particle area in the image block in the grayscale image. The electrostatic adsorption anomaly can be used as an indicator to characterize the degree of noise influence in the image block; further, since electrostatic adsorption will cause the position of particles to be relatively concentrated, while the distribution of noise points is more random and discrete, the noise influence value of the image block is determined based on the distribution and aggregation of the suspected particle area in the image block. The noise influence value can also be used as an indicator to characterize the degree of noise influence in the image block. Finally, the electrostatic adsorption anomaly and the noise influence value of the image block are combined to determine the degree of noise influence of each image block and perform denoising on it, so as to obtain the denoised image corresponding to the grayscale image. At this time, the denoised image can effectively remove noise interference, so the detection of particles in the denoised image can obtain more accurate detection results.
[0156] This embodiment also provides an online detection system for particles on the surface of precision optical components, the system comprising a processor, a memory and a computer program, wherein the memory is used to store the corresponding computer program, the processor is used to run the corresponding computer program, and when the computer program runs on the processor, it can implement any one of the steps of a method for online detection of particles on the surface of precision optical components.
[0157] See also Figure 4 , which shows a system block diagram of an online detection system for particles on the surface of a precision optical element provided by an embodiment of the present invention, including an image acquisition module 401, used to implement step S1 in the above method; an electrostatic adsorption analysis module 402, used to implement step S2 in the above method; a noise impact analysis module 403, used to implement step S3 in the above method; a denoising and detection module 404, used to implement step S4 in the above method.
[0158] See also Figure 5, which shows a schematic diagram of the structure of a computer device provided by an embodiment of the present invention, including: a processor 500, a memory 501, a bus 502 and a communication interface 503, wherein the processor 500, the communication interface 503 and the memory 501 are connected via the bus 502; wherein the memory 501 may include a high-speed random access memory, the bus 502 may be an ISA bus, a PCI bus or an EISA bus, etc., and the processor 500 may be an integrated circuit chip having signal processing capabilities.
[0159] The present invention also provides a computer-readable storage medium corresponding to the method provided in the above-mentioned embodiment. Figure 6 , the storage medium shown is a CD, on which a computer program (ie, a program product) is stored. When the computer program is run by a processor, it will execute the method provided by any of the aforementioned embodiments.
[0160] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), read-only memory (ROM), and other optical and magnetic storage media, which are not listed here one by one.
[0161] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0162] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
Claims
1. A method for online detection of particles on the surface of a precision optical component, characterized in that: The method comprises: Acquire a grayscale image of the surface of the optical element to be measured when it is working, and divide the grayscale image into a plurality of image blocks; Determine the suspected particle area according to the grayscale distribution characteristics of the pixel points in the grayscale image; analyze the position distribution characteristics of the suspected particle area in the image block, the area of the suspected particle area and the position of the image block in the grayscale image to determine the electrostatic adsorption anomaly of each image block; Based on the aggregation of suspected particle areas in the image block, determine the noise impact value corresponding to each image block; Denoising the grayscale image according to the electrostatic adsorption anomaly and the noise influence value of all image blocks to obtain a denoised image; and performing particle detection on the denoised image; The method for obtaining the electrostatic adsorption anomaly comprises: Select any image block as the block to be tested, determine the first electrostatic adsorption factor of the block to be tested based on the area ratio of the suspected particle area in the block to be tested, take the distance between the centroid of the block to be tested and the centroid of the grayscale image as the second electrostatic adsorption factor, and obtain the electrostatic adsorption strength of the block to be tested by combining the first electrostatic adsorption factor and the second electrostatic adsorption factor, and the first electrostatic adsorption factor is positively correlated with the electrostatic adsorption strength, and the second electrostatic adsorption factor is negatively correlated with the electrostatic adsorption strength; Determine an ideal electrostatic adsorption characteristic value of the block to be tested based on the electrostatic adsorption strength of the block and the relative position relationship between the block to be tested and other image blocks; Determine the actual electrostatic adsorption characteristic value of the block to be tested based on the degree of position distribution disorder of the suspected particle area in the block to be tested and the area of each suspected particle area; Analyze the difference between the rational electrostatic adsorption characteristic value and the actual electrostatic adsorption characteristic value of the block to be tested, and obtain the electrostatic adsorption anomaly of the block to be tested; The method for obtaining the noise impact value includes: In any image block, the suspected particle area with the largest area in the image block is used as the reference area, and the remaining suspected particle areas are used as the areas to be analyzed; Determine the clustering factor of the image block according to the difference between the areas of all the regions to be analyzed and the reference area; Determine the noise impact factor of the image block based on the relative position relationship between the region to be analyzed and the reference region and the area of the region to be analyzed; According to the clustering factor and the noise impact factor, a noise impact value of the image block is obtained, and the clustering factor and the noise impact factor both change in the same direction as the noise impact value.
2. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The method for obtaining the ideal electrostatic adsorption characteristic value includes: The image blocks other than the block to be tested are used as interference blocks; The distance between the centroid of the block to be tested and the centroid of each interference block is used as a distance factor; the position weight of each interference block is obtained according to the electrostatic adsorption strength of each interference block and the distance factor corresponding to each interference block, and the electrostatic adsorption strength is positively correlated with the position weight, and the distance factor is negatively correlated with the position weight; Obtain the straight line connecting the center of mass of the block to be tested and the center of mass of each interference block. Taking the horizontal line as the reference, the angle formed by the straight line and the horizontal line in the clockwise direction is used as the relative position eigenvalue of each interference block. Use the position weights of all interference blocks to perform weighted fusion on the corresponding relative position eigenvalues to obtain the ideal electrostatic adsorption eigenvalue corresponding to the block to be tested.
3. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The method for obtaining the actual electrostatic adsorption characteristic value includes: In the block to be tested, a straight line connecting the centroid of the block to be tested and the centroid of each suspected particle matter region is obtained, and the angle formed by the straight line and the horizontal line in the clockwise direction is used as the relative position distribution value of each suspected particle matter region with the horizontal line as a reference; the distance between the centroid of each suspected particle matter region and the centroid of the block to be tested is used as the distance parameter of each suspected particle matter region; The product of the relative position distribution value, distance parameter and area of each suspected particle area is taken as the electrostatic adsorption influence value of each suspected particle area; The actual electrostatic adsorption characteristic value of the block to be tested is obtained by integrating the electrostatic adsorption influence values of all suspected particle areas in the block to be tested.
4. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The method for obtaining the clustering factor includes: The difference between the area of each area to be analyzed and the area of the reference area is used as the area difference factor of each area to be analyzed; The area difference factors of all the regions to be analyzed are averaged to obtain the clustering factor.
5. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The method for obtaining the noise impact factor includes: Obtaining the distance between the centroid of each area to be analyzed and the centroid of the reference area as a distance index of each area to be analyzed; determining an abnormality parameter of each area to be analyzed based on the distance index and area of each area to be analyzed, wherein both the distance index and the area change inversely with the abnormality parameter; All abnormal parameters of the area to be analyzed are averaged to obtain the noise impact factor of the image block.
6. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The method for obtaining the denoised image comprises: The electrostatic adsorption anomaly and the noise influence value of each image block are integrated to obtain an adaptive filter factor of each image block, and the electrostatic adsorption anomaly and the noise influence value are both positively correlated with the adaptive filter factor; The adaptive filtering factor of each image block is used as the scale factor of the Wiener filter, and an adaptive filtering operation is performed on each image block to obtain a denoised image block of each image block; A denoised image corresponding to the grayscale image is obtained based on the denoised image block.
7. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The performing particle detection on the denoised image includes: Perform Canny edge detection on the denoised image to obtain the particle area.
8. The method for online detection of particles on the surface of a precision optical component according to claim 1, characterized in that: The method for obtaining the suspected particle matter area includes: A grayscale image of the surface of the optical element without particles attached is used as a reference image; The grayscale image of the surface of the optical element to be tested is compared with the reference image, and the pixels whose grayscale value difference is greater than a preset grayscale threshold are regarded as suspected particle pixels. A connected domain analysis is performed on the suspected particle pixels in the grayscale image of the surface of the optical element to be tested to obtain the suspected particle area.
Citation Information
Patent Citations
Fixed point noise positioning and sperm microenvironment cleanliness evaluating method
CN108537757A
Lens attached object detection system, terminal and storage medium
CN110544232A
