Sd-oct system for optical element thickness detection and detection method thereof

Abstract

The present application belongs to the field of optical element detection, and particularly relates to a SD-OCT system for optical element thickness detection and a detection method thereof. The detection method comprises mask and Gaussian filtering processing of the OCT image of the optical element to be detected. The peak value point of the OCT image subjected to Gaussian filtering is detected by using a standard deviation peak searching algorithm, and the light intensity mutation position at the film layer interface is identified. The number of film layers is determined by mode analysis of the detected peak value points. The isolated peak value points are filtered out by connected domain analysis of the peak value points subjected to mode analysis, and the missing position is subjected to interpolation processing. The position of each film layer is determined by curve fitting of the peak value points subjected to connected domain analysis, and the actual thickness of the optical element to be detected is calculated. The detection method fuses mask, Gaussian filtering, standard deviation peak searching, mode analysis and connected domain filtering, effectively eliminates the interference of artifacts and noise, and realizes accurate positioning of the film layer boundary.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of optical element detection, and particularly relates to an SD-OCT system for optical element thickness detection and a detection method thereof. BACKGROUND

[0002] With the rapid development of industrial products towards precision and miniaturization, the film layer structure thickness of optical elements such as glass, mobile phone lenses, optical coating, and glasses needs to reach micron-level precision control. The thickness uniformity and accuracy directly determine the optical performance, mechanical stability, and use reliability of the products. For example, if the thickness deviation of the multi-layer coating of a mobile phone lens exceeds the micron level, it will cause imaging distortion; the thickness error of optical glass may cause optical path difference abnormalities, affecting the transmission efficiency of the optical system. Therefore, realizing high-precision and high-efficiency non-destructive detection of optical element film layers has become one of the core needs to ensure the product quality of the optical manufacturing industry.

[0003] Optical coherence tomography (OCT) is a three-dimensional imaging technology based on the Michelson interference principle, which has the advantages of non-contact, high resolution (1-15 μm), high speed, and high signal-to-noise ratio. By using the interference signal of the scattered light at different depths in the sample and the reflected light from the reference arm, the tomographic depth information of the sample can be directly obtained. Combined with two-dimensional galvanometer scanning, three-dimensional structure images can be further generated, realizing micron-level thickness detection and interface positioning, and becoming a potential technical direction to solve the problem of precision detection of optical elements.

[0004] At present, OCT technology has been widely used in biomedical diagnosis (such as retinal imaging) and has achieved mature results. Biological tissues have uniform scattering characteristics, and the gray difference between different tissue layers in the OCT image is obvious, and the boundary is clear. Based on traditional algorithms such as filtering, gradient difference, and multi-scale segmentation, the automatic segmentation and thickness measurement of the multi-layer structure of the retina can be stably completed, showing good robustness and reliability, and laying a foundation for the engineering application of OCT technology.

[0005] Although OCT technology performs excellently in the biomedical field, when it is applied to the detection of industrial samples such as optical elements, due to the differences in sample characteristics and detection requirements, many technical bottlenecks are faced, making it difficult for traditional OCT detection algorithms to be applicable.

[0006] 1. Signal interference problem of industrial samples

[0007] The surface and interior of optical elements (such as glass and optical coating) often have strong reflection and non-uniform scattering: on the one hand, the high reflectivity of the material surface will produce strong noise signals, which will cover the real reflection information of the film layer interface; on the other hand, the non-uniform scattering in the interior will cause the signal intensity of the OCT image to differ significantly at different positions, and obvious burrs will easily appear at the layer boundary, making the interface profile blurred, and the traditional algorithm is easy to misidentify noise as false boundary, or miss detection and missegmentation due to unclear boundary, which seriously affects the detection accuracy.

[0008] 2. Performance limitations of traditional OCT algorithms

[0009] Existing OCT thickness detection algorithms (such as methods based on A-scan scanning) have obvious application scenario limitations:

[0010] Poor adaptability to film layer morphology: only uniform film layers with straight-line distribution can be handled, when the film layer is in a curved shape (such as the arc-shaped coating of a mobile phone lens) or deformed due to stress, the A-scan scanning projects the film layer thickness information into a one-dimensional signal, which cannot reflect the spatial distribution characteristics of the film layer, resulting in significant thickness calculation deviation;

[0011] Insufficient precision and efficiency: most traditional algorithms are designed for millimeter-level thickness samples, and cannot meet the measurement requirements of micrometer-level precision; and some algorithms require manual intervention (such as manually selecting the light intensity peak), which is low in detection efficiency and cannot adapt to batch detection scenarios on production lines;

[0012] Weak anti-interference ability: lack of robust processing mechanism for industrial environment noise, cannot effectively filter isolated noise points or abnormal signals, in the detection of complex multi-layer films (such as 5-layer tight coating of glasses), layer confusion is easy to occur, resulting in unreliable measurement results. SUMMARY

[0013] Therefore, the present application aims to provide an SD-OCT system and a detection method thereof for optical element thickness detection, to solve the technical problem that traditional OCT algorithms cannot be applied to industrial sample detection.

[0014] To achieve the above-mentioned purpose, the technical solution of the present application is as follows:

[0015] An optical element thickness detection method based on an OCT image, realized by an SD-OCT system, comprising the following steps:

[0016] S1: Masking the OCT image of the optical element to be measured to remove non-target areas and retain target areas;

[0017] S2: Gaussian filtering the OCT image after masking to suppress noise in the OCT image;

[0018] S3: Peak value point detection is performed on the OCT image subjected to Gaussian filtering by using a standard deviation peak searching algorithm, and a light intensity mutation position at the interface of the film layer is identified;

[0019] S4: Mode analysis is performed on the detected peak value points to determine the number of film layers;

[0020] S5: Connected domain analysis is performed on the peak value points after mode analysis to filter out isolated peak value points, and interpolation processing is performed on missing positions;

[0021] S6: Curve fitting is performed on the peak value points after connected domain analysis to determine the position of each film layer, and the actual thickness of the optical element to be measured is calculated according to the number of pixels between the film layers and the maximum imaging depth of the SD-OCT system.

[0022] Further, in step S3, the step of performing peak value point detection on the OCT image subjected to Gaussian filtering by using a standard deviation peak searching algorithm includes:

[0023] S31: Calculate the mean value of the gray scale value of each column of pixels in the OCT image and the standard deviation :

[0024] ;

[0025] ;

[0026] wherein, represents the number of effective pixels in the target area, represents the gray scale value of the i-th effective pixel in each column of pixels;

[0027] S32: Based on the mean value of the gray scale value of each column of pixels and the standard deviation , set the adaptive threshold value of the gray scale value of each column of pixels :

[0028] ;

[0029] wherein, represents an adjustable empirical coefficient;

[0030] S33: Each pixel point of the OCT image is traversed column by column, and the pixel point satisfying the following two conditions is marked as a peak value point:

[0031] Condition one: the gray scale value of the pixel point is greater than the adaptive threshold value ;

[0032] Condition two: the gray scale value of the pixel point is greater than the gray scale values of the pixels directly above and below it.

[0033] Further, in step S4, the step of mode analysis on the detected peak points comprises: counting the number of peak points in each column of the OCT image by traversing all columns of the OCT image, and taking the mode of the number of peak points in all columns as the number of the film layers.

[0034] Further, in step S5, the step of connected domain analysis on the peak points after the mode analysis comprises:

[0035] S51: taking each peak point as a center point, setting a fixed size neighborhood for each center point, counting other peak points in the neighborhood except the center point as neighborhood points;

[0036] S52: calculating the proximity of each center point to the neighborhood points

[0037]

[0038] wherein, the center point, the number of neighborhood points in the neighborhood, a certain neighborhood point in the neighborhood, the distance from the center point to the neighborhood point ;

[0039] S53: calculating the average value of the proximities of all peak points and comparing the proximity of each center point with the average value, and filtering out all center points with a proximity less than the average value .

[0040] Further, in step S6, the step of curve fitting on the peak points after the connected domain analysis to determine the position of each film layer comprises:

[0041] S61: setting the number of fitted curves as the mode of the peak points;

[0042] S62: classifying the peak points after the connected domain analysis by using the density clustering method or the distance clustering method;

[0043] S63: fitting a curve on the peak points in each class by using the polynomial method or the spline interpolation method, and the position of each curve in the OCT image is the position of the corresponding film layer.

[0044] Further, in step S6, the calculation formula of the actual thickness of the optical element to be measured is:

[0045]

[0046] wherein, ​​​Pixel number between film layers, Pixel number of camera of spectrometer in SD-OCT system, Maximum imaging depth of SD-OCT system, , Refractive index of optical element to be measured, Central wavelength of light source in SD-OCT system, Resolution of spectrometer, , Working bandwidth of spectrometer.

[0047] Further, in step S3, the minimum interval of the two adjacent peak points is set to 3 pixels.

[0048] An SD-OCT system for optical element thickness detection, comprising:

[0049] A light source for emitting broadband spectrum;

[0050] A fiber coupler for dividing the light emitted by the light source into measurement light and reference light;

[0051] A sample arm and a reference arm for transmitting the measurement light and the reference light, respectively;

[0052] A spectrometer for acquiring an interference signal of the measurement light and the reference light;

[0053] The SD-OCT system further comprises a computer for processing the interference signal acquired by the spectrometer and performing the above-mentioned optical element thickness detection method.

[0054] Compared with the prior art, the present application can achieve the following beneficial effects:

[0055] (1) High precision and high robustness thickness detection is achieved: the algorithm effectively overcomes the interference problems such as strong reflection, non-uniform scattering and image noise of industrial optical samples by a series of processing steps such as mask processing, Gaussian filtering, standard deviation peak finding, mode analysis and connected domain filtering. Experimental results show that the thickness measurement error of the algorithm for glass, mobile phone lens, spectacle lens and other samples can be stably controlled below 2.2%, and the measurement precision of microns can be realized for millimeter to micron level (such as 18 μm) film thickness, which is significantly better than the traditional A-scan scanning method and the row traversal peak finding method;

[0056] (2) The adaptability and automation level of complex situations are improved: the algorithm innovatively uses column direction standard deviation peak searching and peak value quantity mode analysis, which can automatically identify and exclude abnormal data columns, effectively avoiding the interlayer overlapping and misjudgment problems of traditional row traversal method when the film layer distance is too close or presents a curve distribution. Combined with connected domain filtering technology, it can intelligently remove isolated noise points, greatly reducing the need for manual intervention, and realizing full-automatic and high-reliability segmentation and positioning of various film layer morphologies such as straight lines and curves;

[0057] (3) Breakthrough the multi-layer synchronous detection and visualization of single scanning: Unlike the inefficient mode of traditional method of layer-by-layer or point-by-point analysis, the algorithm can simultaneously extract all film layer interfaces in the sample based on single OCT scanning data, and intuitively display the accurate position and morphology of each layer through curve fitting. This not only improves the detection efficiency by several times, but also realizes the visualization of thickness distribution, providing a direct basis for analyzing film layer uniformity, stress deformation, etc.

[0058] (4) Expand the industrial application boundary of OCT technology: This research successfully applies the high-resolution advantage of OCT technology to the detection of weak scattering and multi-layer dense optical elements such as mobile phone lenses and glasses, solving the pain points of traditional technologies such as ultrasonic and spectrum in this scenario, such as insufficient resolution or lack of spatial positioning capability, providing a new and powerful solution for non-destructive testing and quality control of industrial optical products. BRIEF DESCRIPTION OF DRAWINGS

[0059] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification. The illustrations, together with their description, serve to explain the application without unduly limiting it. In the drawings:

[0060] Figure 1 Structure schematic diagram of the SD-OCT system for optical element thickness detection described in the embodiments of the present application;

[0061] Figure 2 Flowchart of the optical element thickness detection method based on OCT image described in the embodiments of the present application;

[0062] Figure 3 Schematic diagram of the original OCT image of the optical element to be tested described in the embodiments of the present application;

[0063] Figure 4 Schematic diagram of the OCT image after mask and Gaussian filtering processing described in the embodiments of the present application;

[0064] Figure 5 Schematic diagram of the OCT image after peak point retrieval described in the embodiments of the present application;

[0065] Figure 6 An OCT image filtered by mode seeking and connected domain filtering according to an embodiment of the present application;

[0066] Figure 7 An illustration of thickness information and layering result of three pieces of glass according to an embodiment of the present application;

[0067] Figure 8 An illustration of thickness information and layering result of a mobile phone lens according to an embodiment of the present application;

[0068] Figure 9 An illustration of thickness curve of a mobile phone lens according to an embodiment of the present application;

[0069] Figure 10 An illustration of thickness information and layering result of a spectacle lens according to an embodiment of the present application;

[0070] Figure 11 An illustration of comparison between column traversal peak seeking result and row traversal peak seeking result according to an embodiment of the present application;

[0071] Figure 12 An illustration of thickness information obtained by a traditional A-scan scanning method.

[0072] The figure mark explanation: light source 1, fiber coupler 2, circulator 3, measurement arm 4, collimating mirror 41, X-axis mirror 42, Y-axis mirror 43, scanning objective 44, reference arm 5, collimating lens 51, focusing lens 52, reflector 53, spectrometer 6, image acquisition system 7, computer 8, optical element to be measured 9. DETAILED DESCRIPTION

[0073] In order to make the purpose, technical scheme and advantages of the present application more clear, the present application is further described in detail below in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and do not constitute a limitation on the present application.

[0074] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0075] In the description of the present application, it is to be understood by those skilled in the art that the orientation or positional relationship indicated by the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second" and the like are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features referred to. Therefore, the features defined with "first", "second" and the like can explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0076] In the description of the present application, it should be noted that unless otherwise specified and limited, the terms "assembly", "connection", "connection" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be connected inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0077] The present application will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0078] The present application provides an SD-OCT (Spectral domain OCT, spectral domain optical coherence tomography) system for optical element thickness detection and an optical element thickness detection method realized by using the SD-OCT system. The structure of the SD-OCT system is as shown in Figure 1As shown, it comprises a light source 1, a fiber coupler 2, a circulator 3, a measurement arm 4, a reference arm 5, a spectrometer 6, an image acquisition system (IMQA) 7 and a computer 8, the measurement arm 4 comprises a collimating mirror 41, an X-axis mirror 42, a Y-axis mirror 43 and a scanning objective 44, the X-axis mirror 42 and the Y-axis mirror 43 form a two-dimensional scanning galvanometer, the reference arm 5 comprises a collimating lens 51, a focusing lens 52 and a reflector 53, the broadband spectrum emitted by the light source 1 is divided into measurement light and reference light by the fiber coupler 2, the reference light becomes parallel light through the collimating lens 51, and then is focused on the reflector 53 through the focusing lens 52, the reference light is reflected by the reflector 53 back to the fiber coupler 2, the measurement light is collimated into parallel light by the collimating mirror 41, and then is deflected by the two-dimensional scanning galvanometer, and then is focused on the optical element 9 to be measured by the scanning objective 44, and the optical path information of the upper surface and the lower surface of different regions of the optical film layer of the optical element 9 to be measured is brought back to the fiber coupler 2, the sample light modulated by the optical film layer of the optical element 9 to be measured and the reference light produce interference signals in the fiber channel, the interference signals are acquired by the spectrometer 6, and the interference signals are acquired by the image acquisition system 7 and sent to the computer 8, and the interference signals acquired by the spectrometer 6 are processed by the computer 8 to detect the thickness of the optical element 9 to be measured.

[0079] The key performance indicators of the SD-OCT system include that the center wavelength of the light source 1 is 840 nm, the bandwidth of the spectrometer 6 is 80 nm, the theoretical axial resolution in air is 3.88 μm, the pixel size of the 2D camera in the spectrometer 6 is 2048×2048, and the scanning speed is 70 kHz, so that excellent tomographic capability is achieved.

[0080] As shown in the figure, the method for detecting the thickness of the optical element specifically comprises the following steps: Figure 2

[0081] S1: Mask processing is performed on the OCT image of the optical element to be measured, non-target regions are removed, and target regions are retained.

[0082] In the actual scanning image, the top edge region and the side edge region of the OCT image of the optical element to be measured may contain artifacts, such as device frames, fixed supports or top marks, as shown in the figure, these structures will produce strong edges similar to bright lines, but are not target bright lines, which will interfere with the detection result, and the optical element to be measured actually needs to be detected is usually not in these non-target regions, so it is necessary to mask process the non-target regions of the OCT image to be black, to exclude the interference of the non-target regions, so that the detection is focused on the target region, and the detection result is prevented from being affected. Figure 3

[0083] S2: Gaussian filtering is performed on the OCT image after the mask processing to suppress the noise in the OCT image.

[0084] ​​Gaussian filtering is a method widely used in image processing, which can suppress noise and preserve image edges and details. The most common role is to eliminate Gaussian noise and salt and pepper noise in the image. By weighted average of the pixel neighborhood, the isolated, abnormal intensity noise points are fused into the surrounding area, so that their intensity tends to approach the surrounding pixels, thereby achieving the effect of smoothing noise. The OCT image after mask processing needs to be denoised before finding the peak point.

[0085] The key of thickness measurement is to accurately identify the positions of the upper and lower surfaces of the film layer. The core advantage of Gaussian filtering is to eliminate false edges and avoid misidentification of noise points as edge points; connect broken edges, smooth adjacent pixels, and help connect the real boundaries broken by noise. It can make the edge more continuous and the positioning more accurate, reducing the measurement jump error.

[0086] Since the mask processing and Gaussian filtering processing are prior art, the specific process will not be described in this application. The OCT image after mask and Gaussian filtering processing is shown in Figure 4 .

[0087] S3: The standard deviation peak finding algorithm is used to detect the peak points of the OCT image after Gaussian filtering, and the light intensity mutation position at the film layer interface is identified.

[0088] In thickness detection, the existence of the film coating will cause the light intensity to change at the interface. The role of the standard deviation peak finding algorithm is to automatically detect the positions of these characteristic peaks, so as to accurately identify the positions corresponding to the plating layer interface. Finally, the gray scale information in the OCT image is converted into a quantifiable thickness value.

[0089] The standard deviation peak finding algorithm is not a global analysis of the entire OCT image, but an independent processing of each column of pixels (i.e. each A-scan signal) in the OCT image. This is because the noise and signal intensity may change at different positions (different columns) of the image. Processing column by column can achieve adaptive peak detection.

[0090] The steps of using the standard deviation peak finding algorithm to detect the peak points of the OCT image include:

[0091] S31: Calculate the mean value of the gray scale value of each column of pixels in the OCT image and the standard deviation :

[0092] ;

[0093] ;

[0094] wherein, represents the number of effective pixels in the target area, Mean of gray value of all pixels in a column

[0095] Mean of gray value of all pixels in a column The mean of gray value of all pixels in a column represents the overall signal level of the column.

[0096] Standard deviation of gray value of all pixels in a column The standard deviation of gray value of all pixels in a column represents the dispersion of the data relative to the mean. The larger the standard deviation, the greater the fluctuation of the data in the column, and the more likely there is a significant peak point.

[0097] S32: Set adaptive threshold for each column of pixels based on the mean and standard deviation of gray value of all pixels in the column :

[0098] ;

[0099] wherein, is an adjustable empirical coefficient, usually taking 1-3.

[0100] S33: Traverse each pixel point of the OCT image column by column, and find the pixel point that meets the following two peak conditions and mark it as a peak point.

[0101] Peak condition one: the gray value of the pixel point is greater than the adaptive threshold .

[0102] The gray value of the pixel point must be higher than the calculated adaptive threshold , to ensure that the pixel point is a significant signal point.

[0103] Peak condition two: the gray value of the pixel point is greater than the gray values of the pixels directly above and below it.

[0104] The gray value of the pixel point must be greater than the gray values of the adjacent pixel points directly above and below it, to ensure that the pixel point is a peak top, not a rising or falling edge.

[0105] The present application sets different thresholds for each column of pixel points, and identifies the pixel points that meet the peak conditions column by column by setting dynamic thresholds, to achieve adaptive peak detection.

[0106] To improve reliability, the standard deviation peak finding algorithm can also add other constraint conditions:

[0107] Minimum peak height: the adaptive threshold described above can be directly used .

[0108] Minimum peak distance: Prevent a wide bright line (spanning multiple pixels) from being identified as multiple closely adjacent peak points. Set the minimum distance between two adjacent peak points to 3 pixels, ensuring that even if the film layer interface has a slight width, it will only be identified as a single peak point.

[0109] Significance: That is, the gray level difference between the current peak point and the adjacent peak point. Further requirements can be made on how much a peak needs to stand out from the valley on both sides to avoid false detection of wide shoulder peaks.

[0110] If a column of pixels has no signal at all (all noise or black background), its standard deviation will be close to 0. When the standard deviation of a column of pixels is detected to be less than a very small value (e.g. 0.001), the standard deviation peak finding algorithm can skip the peak finding of that column of pixels, considering that the column of pixels has no valid peak, thus avoiding false detection.

[0111] The standard deviation peak finding algorithm can be implemented in two modes:

[0112] Mode 1 (automatic mode): Use the mature findpeaks function (present in scientific computing libraries such as MATLAB, Python SciPy, etc.), and pass in adaptive threshold, minimum distance, etc. parameters, for fast automatic peak finding. In most cases, this mode can successfully detect.

[0113] Mode 2 (manual mode): If the findpeaks function cannot find peak values that meet the strict conditions in some columns (possibly due to poor signal), fallback to a more basic manual loop traversal method, i.e. directly apply the logic of step S33 above, and judge the peak value pixel by pixel, to ensure that the standard deviation peak finding algorithm is usable in extreme cases. The OCT image after peak point detection is shown in Figure 5 .

[0114] S4: Perform mode analysis on the detected peak points to determine the number of film layers.

[0115] Mode is an important role in data reliability evaluation and process stability judgment, and can improve the efficiency of judgment when processing multiple measurement results. It does not depend on extreme values and can automatically avoid incorrect data that deviates from the whole. After traversing each column and finding the peak position of each column, record these peak points and mark their horizontal and vertical coordinates with red dots. Take the mode of each column peak point, and the mode is the number of film layers. Columns that do not meet the mode are marked as abnormal columns, and the peak points of abnormal columns are not recorded. In this way, we can get the peak points of each column that meet the requirements, and remove the pseudo-peak points caused by noise.

[0116] ;​

[0117] wherein, represents a column, represents a mode column, represents a peak point, only the column peak point satisfying the mode column is recorded.

[0118] S5: performing connected domain analysis on the peak points after mode analysis, filtering out isolated peak points, and performing interpolation processing on missing positions.

[0119] After finding many peak points, since the extracted thickness information has obvious burrs between the two film layers, the peak points will have some isolated discrete points, at which time the connected domain method is used to traverse all connected regions to filter out isolated discrete points.

[0120] The step of performing connected domain analysis on the peak points after mode analysis includes:

[0121] S51: taking each peak point as a center point, setting a fixed size neighborhood for each center point, counting other peak points in the neighborhood except the center point as neighborhood points.

[0122] S52: calculating the proximity of each center point to neighborhood points :

[0123] ;

[0124] wherein, represents a center point, represents the number of neighborhood points in the neighborhood, represents a neighborhood point in the neighborhood, represents the distance from the center point to the neighborhood point .

[0125] The proximity is used to measure the closeness or aggregation degree between a peak point and its surrounding peak points. High density means that there are many peak points around this peak point, which are closely aggregated together, which is likely to represent a real film layer boundary. Low density means that there are few or no peak points around this peak point, which is likely to be an isolated false peak point caused by random noise.

[0126] S53: calculating the average value of the proximity of all peak points , and comparing the proximity of each center point with the average value, filtering out all center points with proximity less than the average value .

[0127] By the above connected domain analysis, the isolated peak points are removed, and the positions of the removed peak points form the missing positions. At this time, interpolation processing (such as linear interpolation, spline interpolation) is needed for the missing positions to complete the missing positions with the proper peak point coordinates, so as to form a complete and smooth curve, preventing the missing positions from affecting the data during curve fitting.

[0128] The OCT image after searching for the mode and filtering the isolated peak points by the connected domain is shown in FIG. 3. Figure 6

[0129] S6: Curve fitting is performed on the peak points after the connected domain analysis to determine the positions of each film layer, and the actual thickness of the optical element under test is calculated according to the number of pixels between the film layers and the maximum imaging depth of the SD-OCT system.

[0130] The step of performing curve fitting on the peak points after the connected domain analysis to determine the positions of each film layer includes:

[0131] S61: The number of fitting curves is set to the mode of the peak points.

[0132] The mode value obtained by the mode analysis directly corresponds to the number of film layers, and the number of film layers is the number of curves to be fitted. That is, the number of film layers is equal to the number of fitting curves.

[0133] S62: The density clustering method or the distance clustering method is used to classify the peak points after the connected domain analysis.

[0134] After the connected domain filtering, the remaining peak points are effective points of the real film layer interfaces. Since the peak points of the same film layer interface have similar spatial distribution trends in the OCT image, the coordinates (horizontal and vertical coordinates of the image) of the peak points can be used to classify the peak points that are spatially adjacent and have consistent distribution trends into a class by the density clustering (DBSCAN, etc.) or distance clustering method.

[0135] S63: The polynomial method or the spline interpolation method is used to perform curve fitting on the peak points of each class, and the position of each curve in the OCT image is the position of the corresponding film layer.

[0136] Each point set represents incomplete and discrete position information of a film layer. A continuous boundary line needs to be obtained from the point set. A fitting algorithm is used to independently perform curve fitting on each point set, and common fitting algorithms include polynomial, spline interpolation, etc.

[0137] After the fitting is completed, the position of each film layer is no longer an isolated point, but a continuous curve defined by a function. The area passed by the curve in the OCT image is the position of the interface of a certain film layer.

[0138] ​Actual thickness of the optical element to be measured The calculation formula is:

[0139] ;

[0140] Wherein, represents the number of pixels between the film layers, represents the number of pixels between the film layers, represents the maximum imaging depth of the SD-OCT system, , represents the refractive index of the optical element to be measured, represents the central wavelength of the light source, represents the resolution of the spectrometer, , represents the working bandwidth of the spectrometer.

[0141] The detection method provided by the present application can find the film layer and position of the optical element to be measured in one scan. In order to verify the effect of the present application, experiments are carried out on glass, mobile phone lens and glasses respectively.

[0142] Experiment 1: three pieces of glass

[0143] As shown in Figure 7 , (a), (c), (e) in the figure are the original thickness information diagrams of the three pieces of glass, which are roughly black background with some bright lines in the middle, which are the positions of the film layers, and the distance of the bright lines is the thickness of the film layer. In addition to the required bright lines, there are some noise points and protruding burrs in the figure, which need to be removed by algorithm to eliminate the influence of noise and burrs, and find the number and position of the film layers.

[0144] The thickness information and layering results of the first piece of glass are shown as (a), (b) in Figure 7 , the vertical pixel number in (a), (b) is 2048, and the distance between the two lines is 479.6528 pixels; the thickness information and layering results of the second piece of glass are shown as (c), (d) in Figure 7 , the vertical pixel number in (c), (d) is 1024, and the distance between the two lines is 752.118 pixels; the thickness information and layering results of the third piece of glass are shown as (e), (f) in Figure 7The distance between the two lines is 994.166 pixels. After calculating the number of pixels between the two lines, the actual thickness of the three pieces of glass can be calculated. The SD-OCT system can generate multiple thickness information pictures of the same piece of glass in one scan, and the number of pixels between the film layers is the average of multiple pictures. Table 1 below is the model, refractive index, theoretical value, measured value and error analysis of the glass. The refractive index is obtained from the Cauchy dispersion formula, the error value is the difference between the theoretical value and the measured value of the sample, and the error percentage is the ratio of the error value to the theoretical value multiplied by 100%.

[0145] Table 1 Detection results and error analysis of three pieces of glass

[0146]

[0147] As can be seen from Table 1, the detection accuracy of the glass film layer position and thickness is high, with an error of less than 1%, and the error value is in the micrometer level, so the film layer position and thickness of the glass can be accurately found.

[0148] Experiment 2: Mobile phone lens

[0149] The thickness information collected by the spectrometer is two bright lines with a long distance and approximately parallel straight lines, so it is relatively easy to find the position of the film layer and use colored lines to fit, so as to calculate the thickness of the film layer. The following experiment is carried out on a complex mobile phone lens.

[0150] The thickness information and layering results of the mobile phone lens are shown in Figure 8 (a) and (b).

[0151] The thickness curve of the mobile phone lens is shown in Figure 9 (a), and the comparison results of the thickness curve of the mobile phone lens and the theoretical curve are shown in Figure 9 (b). The abscissa of (a) and (b) is the number of horizontal pixels, and the ordinate is the vertical pixel difference between the two fitted curves. The film layer of the mobile phone lens is a smooth film, and the difference between the film layer curve and the theoretical curve is small, so the thickness curve of the mobile phone lens film layer can be accurately drawn, and therefore the detection method is also applicable to the detection of mobile phone lenses.

[0152] Experiment 3: spectacle lens

[0153] The thickness of the spectacle lens is very small (micrometer level), and the thickness information and layering results of the spectacle lens are shown in Figure 10 (a) and (b). As can be seen from Figure 10 (b), the detection method can accurately find the position of the film layer of the spectacle lens, realize the layering and visualization of the film layer of the spectacle lens.

[0154] Table 2 below shows the lens type, refractive index, theoretical value, measured value, and error analysis.

[0155] Table 2. Test results and error analysis of spectacle lenses

[0156]

[0157] In Table 2, the first column (1-2) represents the film thickness between the first and second layers, (2-3) represents the film thickness between the second and third layers, (3-4) represents the film thickness between the third and fourth layers, and (4-5) represents the film thickness between the fourth and fifth layers. As shown in Table 2, this testing method is also applicable to the testing of spectacle lenses. The measurement error between the theoretical and actual values ​​is very small, below 2.2%, thus allowing for accurate location and thickness determination of the film layers in spectacle lenses.

[0158] It should be noted that, assuming the remaining steps of the method remain unchanged, when finding the peak point, if we iterate through each row of pixels and find the peak point by calculating the average brightness of each row of pixels and then finding the row with the highest brightness, the fitted result will be as follows when processing two closely spaced bright lines: Figure 11 As shown, Figure 11 Image (a) in the image is the correctly identified image. Figure 11 (b) in the figure shows the error identification. As can be seen, because the distance between the two film layers is too close, and the two film layers are roughly distributed diagonally from the upper left to the lower right, there is a section of overlap between the lower right end of the upper bright line and the upper left end of the lower bright line in the row direction. If the peak is searched by traversing each row, the overlapping part will be incorrect, which will have a great impact on the fitted curve results and the measurement results, making it difficult to find the correct film layer position and calculate the correct thickness value.

[0159] The edge tracking method involves setting a threshold to convert the denoised OCT image into a binary image, and then selecting the first pixel with an intensity of 1 on the right side of the brightness as the starting point for edge tracking. This method is prone to errors when the distance between the two membranes is too close, or when there are burrs connecting the two membranes. Figure 11 The case shown in (b) is as follows.

[0160] Traditional methods for detecting thickness using OCT systems employ A-scan scanning to project the sample's thickness information, such as... Figure 12As shown, several points with the highest light intensity are found, thereby converting the thickness information into one-dimensional intensity values, subtracting the abscissa of the two light intensity peaks to obtain the pixel difference, and then obtaining the thickness value through the system imaging depth and the sample refractive index. Compared with the algorithm fitting curve, the error of this method is larger, usually in the order of several microns, and the position of the film layer cannot be directly displayed, so the thickness visualization cannot be realized. Moreover, this method needs to be manually calculated, and the efficiency is low. When the sample film layer is subjected to stress or the film layer itself is curved, this method cannot find the bending degree of the film layer, and the influence of stress on the film layer cannot be judged.

[0161] The present application combines mask processing, Gaussian filtering, standard deviation peak searching, mode analysis and connected domain filtering, and constructs a complete automatic detection process, effectively suppresses the noise interference caused by strong reflection and scattering unevenness, and realizes high-precision positioning of the film layer boundary. Compared with ultrasonic wave (resolution millimeter level) and spectral reflection method (lack of spatial resolution), the detection method based on OCT image provides thickness distribution with millimeter level range and micrometer level precision, which is more suitable for weak scattering material detection. Compared with the line traversal method, the present method avoids overlapping error when the film layers are close to each other. The line traversal method is prone to measurement deviation due to overlapping parts when the bright lines are distributed obliquely. Compared with the traditional OCT thickness detection method, the present method can realize thickness visualization and directly display the position of the film layer, and can handle the case where the film layer is a curve, and is more accurate and efficient. The present method suppresses noise points by introducing Gaussian filtering, and excludes non-target area interference by mask processing; the standard deviation peak searching and mode analysis accurately identify the film layer position and number; the connected domain filtering further eliminates abnormal points, and finally the thickness value is obtained through curve fitting. Experimental results show that the present method can extract all film layer thicknesses through single scanning in typical samples such as glass, mobile phone lenses and glasses, the detection efficiency is significantly improved, the measurement precision reaches micrometer level, the error is controlled below 2.2%, which is better than ultrasonic wave and spectral reflection methods. The present method not only expands the application range of OCT technology in weak scattering and multi-layer optical material thickness detection, but also confirms the core advantages of OCT in industrial nondestructive testing.

[0162] It should be understood that the various forms of flow shown above can be reordered, steps added or deleted. For example, the steps described in the present disclosure can be executed in parallel, in sequence, or in a different order, as long as the desired results of the technical solutions of the present disclosure can be achieved, which is not limited herein.

[0163] The above detailed description does not limit the scope of the application. Various modifications, combinations, sub-combinations and alternatives can be made to the detailed description. Any modification, equivalent replacement and improvement etc. made within the spirit and principle of the application shall be included in the scope of the application.

Claims

1. An OCT image-based optical element thickness detection method, implemented using an SD-OCT system, characterized in that, The method comprises the following steps: S1: Mask processing is performed on an OCT image of the optical element to be measured, non-target regions are removed, and target regions are retained; S2: Gaussian filtering is performed on the OCT image after the mask processing to suppress noise in the OCT image; S3: Peak point detection is performed on the OCT image after the Gaussian filtering using a standard deviation peak searching algorithm to identify light intensity mutation positions at film layer interfaces; S4: Mode analysis is performed on the detected peak points to determine the number of film layers; S5: Connected domain analysis is performed on the peak points after the mode analysis to filter out isolated peak points and perform interpolation processing on missing positions; S6: Curve fitting is performed on the peak points after the connected domain analysis to determine the positions of each film layer, and the actual thickness of the optical element to be measured is calculated according to the number of pixels between the film layers and the maximum imaging depth of the SD-OCT system.

2. The OCT image-based optical element thickness detection method according to claim 1, characterized by, In step S3, the step of performing peak point detection on the OCT image after the Gaussian filtering using the standard deviation peak searching algorithm comprises: S31: Calculate the mean value of the gray value of each column of pixels in the OCT image and standard deviation : ； ； wherein, represents the number of valid pixels within the target region, represents the gray value of the i-th valid pixel within each column of pixels; S32: mean value of each column pixel gray value and standard deviation set adaptive threshold of each column pixel gray value : ； wherein represents an adjustable empirical coefficient; S33: Each pixel point of the OCT image is traversed column by column to find pixel points satisfying the following two conditions, and the pixel points are marked as peak points: Condition one: the gray value of the pixel point is greater than the adaptive threshold ; Condition two: The gray value of the pixel point is greater than the gray values of the pixels directly above and below it.

3. The OCT image-based optical element thickness detection method according to claim 1, characterized by, In step S4, the step of performing mode analysis on the detected peak points comprises: by traversing all columns of the OCT image, the number of peak points in each column is counted, and the mode of all column peak points is taken as the number of film layers.

4. The OCT image-based optical element thickness detection method according to claim 1, characterized by, In step S5, the step of performing connected domain analysis on the peak points after the mode analysis comprises: S51: Each peak point is taken as a center point, a fixed-size neighborhood is set for each center point, other peak points in the neighborhood except the center point are counted and taken as neighborhood points; S52: Calculate the proximity of each center point to the neighborhood points : ； wherein, denotes the center point, denotes the number of neighborhood points within the neighborhood, denotes a certain neighborhood point within the neighborhood, denotes the center point to a neighborhood point distance; S53: Calculate the average of the proximity of all peak points and compare all proximities to the proximity of each center point, filter out the center points whose proximity is less than the average value.

5. The OCT image-based optical element thickness detection method according to claim 1, wherein In step S6, the step of performing curve fitting on the peak points after the connected domain analysis to determine the positions of each film layer comprises: S61: The number of fitting curves is set to the mode of the peak points; S62: The density clustering method or the distance clustering method is used to classify the peak points after the connected domain analysis; S63: The polynomial method or the spline interpolation method is used to perform curve fitting on the peak points of each class, and the position of each curve in the OCT image is the position of the corresponding film layer.

6. The OCT image-based optical element thickness detection method according to claim 1, wherein In step S6, the actual thickness of the optical element to be measured The calculation formula is: ； wherein, represents the number of pixels between the film layers, represents the number of pixels between the film layers, represents the maximum imaging depth of the SD-OCT system, , represents the refractive index of the optical element to be measured, represents the center wavelength of the light source in the SD-OCT system, represents the resolution of the spectrometer, , represents the working bandwidth of the spectrometer.

7. The OCT image-based optical element thickness detection method according to claim 1, wherein In step S3, the minimum spacing between two adjacent peak points is set to 3 pixels.

8. An SD-OCT system for optical element thickness detection, comprising: a light source for emitting a broadband spectrum; a fiber coupler for dividing the light emitted by the light source into measurement light and reference light; a sample arm and a reference arm for transmitting the measurement light and the reference light, respectively; a spectrometer for acquiring interference signals of the measurement light and the reference light; characterized in that the SD-OCT system further comprises a computer for processing the interference signals acquired by the spectrometer and performing the optical element thickness detection method according to any one of claims 1 to 6.

Images