Image drift monitoring method and system for asymmetric spatial heterodyne interferometer

Abstract

The invention discloses an image drift monitoring method for an asymmetric spatial heterodyne interferometer. The method comprises the following steps: constructing a scale pattern; acquiring a first interference image and a second interference image; a signal selection and interpolation step: selecting a plurality of rows of pixels with a scale pattern in the interference image for averaging, and taking the selected pixels as one-dimensional interference image data containing the scale pattern; calculating the characteristic amplitude of the edge of the scale pattern; a scale pattern feature matching step: aligning left and right feature edges of each scale pattern, selecting points with the same amplitude on the left and right edges of each scale pattern, and subtracting horizontal coordinates of the left and right edges under the same amplitude to obtain feature widths W1 and W2 of the scale patterns of the first interference image and the second interference image; comparing the W1 with the W2, and if a difference value between the W1 and the W2 is smaller than a set number of pixels, finishing feature matching; respectively calculating the average values of the horizontal coordinates of the left and right edges of the scale patterns of the first and second interference images after feature matching, wherein the difference between the average values is the horizontal drift distance of the two sampling images.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to an image drift monitoring method and system for an asymmetric spatial heterodyne interferometer. Background Technology

[0002] Asymmetric spatial heterodyne interferometers have wide applications in the detection of neutral winds in the middle and upper atmosphere. This instrument generates interference fringes by directly receiving spectral lines emitted from a target light source. By calculating the phase change of the interference fringes, the Doppler frequency shift of the target light source can be obtained, and thus the radial velocity of the target can be calculated. In actual detection, the asymmetric spatial heterodyne interferometer can detect the neutral wind field at a corresponding altitude by detecting airglow at a specific wavelength.

[0003] Interference fringes generated by asymmetric spatial heterodyne interferometers are highly sensitive to the surrounding environment. Changes in ambient temperature cause the interferometric image to drift. Although this drift may only be at the sub-pixel level, it still introduces significant errors into the velocity inversion results. To correct for these errors, an effective method capable of monitoring interferometric image drift in real time is needed.

[0004] Traditional techniques for monitoring interferometric image drift primarily rely on grating etching. This process can produce highly precise grating scales with clear outlines. Subsequently, images containing and without scale signals are processed separately to calculate relevant parameters of the scale signal and fit a functional expression for the scale signal, thereby achieving precise monitoring of the scale's horizontal displacement. For example, patent document CN111089537B discloses a technique entitled "A method and system for detecting the imaging position shift of a Doppler differential interferometer, overcoming the influence of Doppler differential interferometer image plane drift on the accuracy of inversion phase, to obtain more accurate phase results. Periodic, uniform grooves are etched on the grating diffraction surface of the Doppler differential interferometer as a scale, resulting in periodic shadow patterns corresponding to the groove positions in the fringe image formed by the interferometer. Based on a fitting method, the edges of the shadow patterns are finely detected with sub-pixel precision to obtain the pixel position of the edge center. The drift of the interferometer imaging position is determined by the change in the edge center position. The inversion phase is corrected based on the overall offset of the detected image on the detector image plane, resulting in an accurate inversion phase." However, the above methods place extremely stringent requirements on the precision of optical processing, typically requiring micrometer- to nanometer-level accuracy. If the processing precision is insufficient, for example, when laser marking technology can only provide millimeter- to micrometer-level accuracy, problems such as inconsistent pattern size or spacing, flaws at the pattern edges, and uneven distribution of reflectivity within the pattern may make it difficult to achieve accurate fitting and monitoring. Summary of the Invention

[0005] The purpose of this application is to monitor image drift of an asymmetric spatial heterodyne interferometer and correct the wind measurement deviation caused by image drift, especially when the scale processing accuracy is low. It provides an image drift monitoring method and system for an asymmetric spatial heterodyne interferometer, which corrects the inversion phase based on the overall offset of the monitored image on the detector image plane, eliminates the influence of drift on the accuracy of the inversion phase, and obtains more accurate phase results.

[0006] To achieve the above objectives, this application provides an image drift monitoring method for an asymmetric spatial heterodyne interferometer, characterized by comprising: The step of constructing the scale pattern is to construct a periodic scale pattern in the edge region of the grating of the asymmetric spatial heterodyne interferometer. The image acquisition step involves acquiring the first and second interferometric images formed by the asymmetric spatial heterodyne interferometer, respectively. The signal selection and interpolation steps involve averaging multiple rows of pixels with scale patterns in the interferometric image to obtain one-dimensional interferometric image data containing the scale pattern. The horizontal axis represents the position of the pixel, and the vertical axis represents the amplitude of the image. Linear interpolation is then performed on the one-dimensional interferometric image data to expand the data volume. The step of extracting scale pattern features involves calculating the feature amplitude of the scale pattern edge for the one-dimensional interferometric image data. The step of matching scale pattern features involves aligning the left and right feature edges of each scale pattern, selecting points with the same amplitude on the left and right edges of each scale pattern, and subtracting the horizontal coordinates of the left and right edges under the same amplitude to obtain the feature width W1 of the scale pattern in the first interference image and the feature width W2 of the scale pattern in the second interference image; comparing W1 and W2, if the difference between the two is less than a set value of pixels, then feature matching is complete. In the step of calculating the drift, the average value of the horizontal coordinates of the left and right edges of the scale pattern in the first and second interferometric images after feature matching is calculated respectively, and the difference between the two is the horizontal drift of the two sampled images.

[0007] As an improvement to the above method, the step of extracting scale pattern features further includes: selecting data points in the scale signal that are at the feature amplitude, expanding them on the left and right sides respectively, selecting more additional data points, and performing grouping operations based on the continuity of their position coordinates. The obtained amplitude and position relationship information is the extracted scale pattern feature.

[0008] As another improvement to the above method, the step of extracting the scale pattern features further includes: checking the accuracy of the expanded and grouped data, based on the monotonicity of the data and the positional interval of the average amplitude of each group; if a group of data is non-monotonic, or the positional interval exceeds a specific range of the scale period, the data grouping is determined to be incorrect; then the expansion distance is reduced, and the grouping operation is repeated until the grouping result passes the check or the image is determined to be abnormal and unable to extract effective features.

[0009] As another improvement to the above method, in the step of extracting the scale pattern features, the calculation of the feature amplitude of the scale pattern edge includes extracting the peak and valley values ​​from the one-dimensional interference image signal, calculating the average value M of the peak value and the average value N of the valley value, and using (M+N) / 2 as the feature amplitude of the scale pattern edge.

[0010] As an improvement to the above method, before the signal selection and interpolation step, an abnormal pixel correction step is included, in which abnormal peak pixels with higher signal strength than the surrounding pixels are identified by a preset threshold in the first and second interference images, and the abnormal peak pixels are replaced by the average value of the adjacent pixels in the same column.

[0011] As a further improvement to the above method, the method for determining the preset threshold includes: dividing the image into several sub-blocks, calculating the independent mean amplitude and standard deviation within each sub-block, and setting the threshold of the i-th sub-block as follows: ,in It is the local pixel mean. It is the standard deviation, and k is a constant ranging from 3 to 5.

[0012] As a further improvement to the above method, in the signal selection and interpolation step, the data volume is increased by 1000 times; in the matching scale pattern feature step, the set value is 1 / 1000.

[0013] As a further improvement to the above method, in the step of matching the scale pattern features, aligning the left and right feature edges of each scale pattern specifically includes: for the i-th scale pattern of the first interferometric image, generating a unified amplitude axis based on a set step size, and then mapping the original discrete amplitudes GL1(i) and GR1(i) on both sides of the left and right edges of the scale pattern onto the unified amplitude axis using a linear interpolation method, and finally outputting the aligned left and right edge data UL1(i) and UR1(i); for the i-th scale pattern of the second interferometric image, generating a unified amplitude axis based on a specified step size, and then mapping the original discrete amplitudes GL2(i) and GR2(i) on both sides of the left and right edges of the scale pattern onto the unified amplitude axis using a linear interpolation method, and finally outputting the aligned left and right edge data UL2(i) and UR2(i). The step of matching scale pattern features further includes: calculating the feature width W1(i) of each scale pattern with the same amplitude based on UL1(i) and UR1(i); calculating the feature width W2(i) of each scale pattern with the same amplitude based on UL2(i) and UR2(i); comparing the feature widths contained in W1(i) with the feature widths contained in W2(i) one by one, and if there is a set of feature widths with a difference less than the set value of pixels, then the feature matching of the i-th scale is considered to be completed; the position coordinates of the first interference image scale pattern and the second interference image scale pattern that satisfy this condition are UL1(i)', UR1(i)' and UL2(i)', UR2(i)'; The step of calculating the drift amount specifically includes: calculating the average values ​​of UL1(i)' and UR1(i)' and the average values ​​of UL2(i)' and UR2(i)' of multiple scale patterns respectively. The difference between the average values ​​of UL1(i)' and UL2(i)' and the difference between the average values ​​of UR1(i)' and UR2(i)' are the horizontal drift amounts of the first and second interferometric images.

[0014] To achieve the above objectives, this application also provides an image drift monitoring system for an asymmetric spatial heterodyne interferometer, including a processor and a memory, characterized in that the memory stores a computer program, and when the computer program is run in the processor, it implements the steps from image acquisition to drift calculation described above.

[0015] To achieve the above objectives, this application also provides a computer-readable storage medium, characterized in that: it stores a computer program, which, when executed by a processor, implements the steps from the image acquisition step to the calculation of the drift amount described above.

[0016] Compared with the prior art, the advantages of this application are: This invention can correct image drift caused by temperature changes during wind measurement in asymmetric spatial heterodyne interferometers, effectively reducing the instrument's measurement error. Compared with traditional techniques, this invention innovatively improves the extraction and matching method of scale pattern features. Instead of the complex operation of fitting functions required in traditional methods, it utilizes the constant width of the same scale feature to accurately match the scale positions obtained from two consecutive samples. This method eliminates the need to construct a high-precision grating scale; it can achieve the goal using only a common commercial laser marking machine, reducing equipment costs and operational complexity. This invention can be used for atmospheric wind field detection, improving the instrument's detection accuracy. Attached Figure Description

[0017] Figure 1 This is an interference image from a specific embodiment of the present invention; Figure 2(a) is a schematic diagram of selecting feature amplitude in the image drift monitoring method of a specific embodiment of the present invention; Figure 2(b) is a schematic diagram of the range selection based on feature amplitude expansion in the image drift monitoring method according to a specific embodiment of the present invention; Figure 2(c) is a schematic diagram of the alignment of the left and right feature edges of the scale pattern in the image drift monitoring method of a specific embodiment of the present invention; Figure 2(d) is a schematic diagram of the width of the matching scale pattern feature in the image drift monitoring method of a specific embodiment of the present invention; Figure 3(a) shows the ambient temperature during the observation period on February 17, 2025, in the image drift monitoring method of the present invention, using an asymmetric spatial heterodyne interferometer to detect the neutral wind field at the top of the intermediate layer in Weihai area. Figure 3(b) is a schematic diagram of the horizontal distance of interferometric image drift in the image drift monitoring method of a specific embodiment of the present invention; Figure 3(c) shows the wind field inversion results before interferometric image drift correction in a specific embodiment of the present invention; Figure 3(d) shows the wind field inversion result after interferometric image drift correction in the image drift monitoring method of a specific embodiment of the present invention; Figure 4 This is an enlarged view of the data of the first scale pattern in Figure 2(b). Detailed Implementation

[0018] The technical solutions provided in this application are further illustrated below with reference to the embodiments.

[0019] In a specific embodiment of this invention, a laser marking machine is used to construct a periodic scale pattern on the edge of a grating in an asymmetric spatial heterodyne interferometer. Features of the scale pattern are extracted using initial screening based on feature amplitude and a series of judgments, and the features of the scale pattern in two consecutive interferometric images are matched. The distance of the horizontal displacement of the feature edges of the two scale patterns is calculated to monitor image drift in the asymmetric spatial heterodyne interferometer. This scheme utilizes the characteristic that the scale feature width remains constant, does not require precise fitting of a function containing the scale signal, has no specific requirements for marking accuracy, and can be implemented using a common commercial marking machine.

[0020] As a specific embodiment of the present invention, an image drift monitoring method for an asymmetric spatial heterodyne interferometer may include the following steps: Step 1: Construct the ruler pattern By using a laser marking machine to vaporize the metal on the surface of a grating in an asymmetric spatial heterodyne interferometer, a periodic scale pattern is constructed in the edge region (e.g., ...). Figure 1 (As shown).

[0021] Step 2: Acquire Images Interference image 1 and interference image 2, formed by an asymmetric spatial heterodyne interferometer, were acquired using a detector.

[0022] Step 3: Abnormal pixel correction Peak pixels with significantly higher signal strength than surrounding pixels are identified by using a preset threshold. Then, the average value of adjacent pixels in the same column is used to replace the signal of these abnormal pixels, smoothing out local anomalies. A method for setting the preset threshold could be a block-based thresholding approach, dividing the image into several sub-blocks (e.g., 64x64), calculating the independent mean and standard deviation within each sub-block, and setting the threshold for the i-th sub-image. ,in It is the local pixel mean. It is the standard deviation, and k is 3 to 5.

[0023] Step 4: Signal Selection and Interpolation In a two-dimensional interferometric image, pixels from n rows (e.g., rows 7, 8, 9, 10, or 11) with a typical scale pattern are averaged to create a one-dimensional interferometric image containing the scale pattern. The horizontal axis represents the pixel position, and the vertical axis represents the image amplitude. To determine sub-pixel-level image drift, the pixel positions and amplitudes of the one-dimensional interferometric image are linearly interpolated, expanding the existing data by a factor of R. The value of R can be relatively large, such as 1000, meaning that 1000 pixel positions and amplitudes are interpolated for each pixel. In the following descriptions of pixels, the pixel position before this expansion will be used; therefore, when describing, for example, n / N pixels, the expanded data volume will be considered.

[0024] Step 5: Extract the features of the ruler pattern Interference images contain both interference fringes and scale patterns. During feature extraction, preliminary screening can be performed using feature amplitudes, followed by a series of judgment processes to eliminate interference from interference fringes, thereby achieving accurate extraction of the scale pattern features.

[0025] Specifically, for the one-dimensional interferometric image signal, the peak and valley values ​​are extracted respectively, and the average value M of the peak value and the average value N of the valley value are calculated. (M+N) / 2 is used as the characteristic amplitude of the scale pattern edge. Data points in the scale signal that are at the characteristic amplitude (as shown in Figure 2(a)) are selected, and a preset distance is extended on both sides of them. More data points are selected to improve accuracy. The initial extension distance is set to 1 pixel (as shown in Figure 2(b)).

[0026] For data points acquired after expansion around a specific amplitude, grouping is performed based on the continuity of their position coordinates. To prevent interference fringe signals from being mistakenly identified as scale edges, the accuracy of the expanded and grouped data needs to be verified. The verification is based on two main criteria: the monotonicity of the data and the positional interval of the average amplitude in each group. If a group of data exhibits non-monotonicity, or the positional interval between adjacent groups exceeds a specific range of the scale period, the data grouping is deemed incorrect. In this case, the expansion distance needs to be reduced, and the grouping operation repeated until the grouping results pass the verification, or the image is determined to be abnormal and unable to extract effective features.

[0027] After grouping the scale signals of interferometric images 1 and 2, the obtained amplitude and position relationship information is the extracted scale pattern feature.

[0028] Step 6: Match the ruler pattern features The amplitude of the scale signal varies across different interferometric images due to factors such as airglow intensity and ambient light intensity. However, for the same scale, the characteristic width of the scale pattern remains constant across each sampled image. Therefore, the horizontal position of the image can be determined by matching based on the width.

[0029] First, the discrete amplitude values ​​of the left and right feature edges of each ruler pattern are aligned to a uniform scale. A uniform amplitude axis is set, and linear interpolation is performed on the feature edges of each ruler to map the ordinate onto this amplitude axis (as shown in Figure 2(c)). Then, points with the same amplitude are selected on the left and right edges of each ruler pattern after alignment. The horizontal coordinates of the left and right edges are subtracted at the same amplitude to obtain the feature widths W1 and W2 of each ruler pattern. W1 and W2 are compared. If the difference between the two is less than 1 / 1000 of a pixel, the feature matching is considered complete (as shown in Figure 2(d)).

[0030] Step 7: Calculate the drift amount Calculate the average value of the ruler edge position of the two interferometric images before and after, and the difference between the two values ​​is the horizontal drift of the two sampled images.

[0031] Figures 3(a), 3(b), 3(c), and 3(d) illustrate a case study of probing the neutral wind field at the mesotope using an asymmetric spatial heterodyne interferometer. In this case, as shown in Figure 3(a), the ambient temperature undergoes a significant change, with the vertical axis representing ambient temperature (°C). As shown in Figure 3(b), the change in ambient temperature causes a drift in the instrument's imaging system, manifested as a horizontal shift in the interferometric image. This alters the phase difference of the interferometric fringes originally caused by changes in Doppler velocity, with the vertical axis representing image drift (in pixels) and the horizontal axis representing time (UTC).

[0032] The following is the specific method used in this case to monitor image drift using the present invention: Step 1: Construct the ruler pattern By using a laser marking machine to vaporize the metal on the surface of a grating in an asymmetric spatial heterodyne interferometer, a periodic scale pattern is constructed in the edge region (e.g., ...). Figure 1 (As shown). In this case, the periodicity of the scale pattern can be constructed as needed, for example, 1mm. Figure 1 It is an interference fringe image acquired by the detector, which contains a scale pattern, appearing as periodic black blocks in the image.

[0033] Step 2: Acquire Images Interference image 1 and interference image 2, formed by an asymmetric spatial heterodyne interferometer, were acquired using a detector.

[0034] Step 3: Abnormal pixel correction Peak pixels with significantly higher signal strength than surrounding pixels are identified by using a preset threshold. Then, the average value of adjacent pixels in the same column is used to replace the signal of these abnormal pixels, smoothing out local anomalies. In this case, the preset threshold uses a block-based thresholding method, dividing the image into several sub-blocks (e.g., 64×64). The mean and standard deviation are calculated independently for each sub-block, and the threshold for the i-th sub-block is set to [value missing]. ,in It is the local pixel mean. It is the standard deviation, and k can be 3 to 5. For example, k=4.

[0035] Step 4: Signal Selection and Interpolation In a two-dimensional interferometric image, approximately 10 rows of pixels with a typical scale pattern are selected and averaged to obtain one-dimensional interferometric image data containing the scale pattern. The horizontal axis represents the pixel position, and the vertical axis represents the image amplitude. To determine sub-pixel level image drift changes, linear interpolation is performed on the one-dimensional interferometric image, expanding the existing data volume by 1000 times.

[0036] Step 5: Extract the features of the ruler pattern For the one-dimensional interferometric image signal, peak and valley values ​​are extracted respectively, and the average value M of the peak value and the average value N of the valley value are calculated. (M+N) / 2 is used as the characteristic amplitude of the scale pattern edge. As shown in Figure 2(a), data points in the scale signal at the characteristic amplitude are selected (the vertical axis of Figure 2(a) represents the detector signal strength (Relative intensity), which is the "amplitude" mentioned above, and the horizontal axis represents the detector pixels. The blue curve represents image 1, the orange curve represents image 2, the green dot A1 represents the characteristic amplitude data point of image 1, and the red pentagram A2 represents the characteristic amplitude data point of image 2). A preset distance is extended to the left and right sides of the characteristic amplitude data point to select more data points to improve accuracy. The initial preset distance is set to 1 pixel. Figure 2b As shown, the vertical axis represents the detector signal strength (Relative intensity), and the horizontal axis represents the detector pixels. The blue curve represents image 1, the orange curve represents image 2, the green dot G1 represents the data point of image 1, and the red pentagram G2 represents the data point of image 2.

[0037] For data points acquired after expansion around a specific amplitude, grouping is performed based on the continuity of their position coordinates. To prevent interference fringe signals from being mistakenly identified as scale edges, the accuracy of the expanded and grouped data needs to be verified. The verification is based on two main criteria: the monotonicity of the data and the positional interval of the average amplitude in each group. If a group of data exhibits non-monotonicity, or the positional interval between adjacent groups exceeds a specific range of the scale period, the data grouping is deemed incorrect. In this case, the preset distance needs to be reduced (e.g., reduced to 2 / 3 of a pixel) for expansion, and the grouping operation needs to be repeated until the grouping results pass the verification, or the image is determined to have anomalies that prevent the extraction of effective features.

[0038] The amplitude and positional relationship obtained after grouping the scale signals in interferometric images 1 and 2 constitute the extracted scale pattern features G1 and G2. Taking G1 in Figure 2(b) as an example, G1 contains 42 sets of scale feature edge coordinates and amplitudes. The 42 sets of coordinates are divided into odd and even numbers, corresponding to the left and right edges of 21 scale patterns, respectively. The left and right feature edges of the i-th scale pattern in interferometric image 1 are represented as GL1(i) and GR1(i), respectively. The left and right feature edges of the i-th scale pattern in interferometric image 2 are represented as GL2(i) and GR2(i), respectively. Step 6: Match the ruler pattern features For the first interferogram 1, the discrete amplitudes of the left and right edges of each scale pattern need to be uniformly scaled. For the i-th scale pattern, a uniform amplitude axis is generated based on a set step size, and then the original discrete amplitudes GL1(i) and GR1(i) on both sides of the left and right edges of the scale pattern are mapped onto this uniform amplitude axis using a linear interpolation method. Finally, the aligned left and right edge data UL1(i) and UR1(i) of the scale pattern are output. Similarly, for the second interferogram 2, the discrete amplitudes of the left and right edges of each scale pattern need to be uniformly scaled. For the i-th scale pattern, a uniform amplitude axis is generated based on a set step size, and then the original discrete amplitudes GL2(i) and GR2(i) on both sides of the left and right edges of the scale pattern are mapped onto this uniform amplitude axis using a linear interpolation method. Finally, the aligned left and right edge data UL2(i) and UR2(i) of the scale pattern are output.

[0039] Figure 4 This is an enlarged view of the first scale pattern in Figure 2(b) (the red dot is GL1(1), and the purple dot is GR1(1)). The vertical axis represents the signal intensity on the detector (Relative intensity), and the horizontal axis represents the pixel position of the detector. Taking the left and right edges of the scale pattern in this figure as an example, assuming there are n+1 data points on each edge, the coordinates of the first and last data points of GL1(1) are denoted as follows: and The indices of the intermediate data point coordinates are represented by 1, 2, 3...n-1; similarly, the coordinates of the first and last data points of GR1(1) are denoted as... and The relationship of the x-axis is as follows .according to Establish a unified amplitude axis, the range of which is The amplitude step size interval is set to p, which is 0.01 in this implementation. Linear interpolation functions are constructed based on the coordinates of GL1(1) and GR1(1): y u In the interval hour, ;y u In the interval hour, Each interpolation function calculates the interpolation result on the uniform amplitude axis according to the set step size. The left and right edges formed by the new interpolation points are continuous data with equal spacing, represented as UL1(1) and UR1(1).

[0040] As shown in Figure 2(c), the vertical axis represents the detector signal strength (Relative intensity), and the horizontal axis represents the detector pixels. The red dot UL1 represents the data point on the left edge of the ruler pattern after image 1 is aligned, and the purple dot UR1 represents the data point on the right edge of the ruler pattern after image 1 is aligned. The blue pentagram UL2 represents the data point on the left edge of the ruler pattern after image 2 is aligned, and the green pentagram UR2 represents the data point on the right edge of the ruler pattern after image 2 is aligned.

[0041] Calculate the feature width W1(i) of each scale pattern with the same amplitude based on UL1(i) and UR1(i). Perform the same processing on the scale feature patterns of interference image 2, and calculate the feature width W2(i) of each scale pattern with the same amplitude based on UL2(i) and UR2(i). Compare the feature widths contained in W1(i) with the feature widths contained in W2(i) one by one. If there is a set of feature widths with a difference of less than a set value (e.g., 1 / 1000) pixels, then the feature matching of the i-th scale is considered complete. The coordinates of the scale patterns of the first interferometric image 1 and the second interferometric image 2 that satisfy this condition are UL1(i)', UR1(i)' and UL2(i)', UR2(i)' (as shown in Figure 2(d), where the vertical axis represents the detector signal intensity and the horizontal axis represents the detector pixels. The red dots UL1' and UR1' represent data points with the same amplitude on the left and right edges of the scale pattern after matching in image 1, and the blue pentagrams UL2' and UR2' represent data points with the same amplitude on the left and right edges of the scale pattern after matching in image 2). In this case, W1(i) and W2(i) will contain thousands of data points. If a mismatch occurs, it indicates that the image noise is severe and image drift cannot be effectively detected. The average drift of the previous and next images can be used to estimate the current image drift.

[0042] Step 7: Calculate the drift amount The average values ​​of UL1(i)' and UR1(i)' and the average values ​​of UL2(i)' and UR2(i)' for multiple scale patterns are calculated respectively. The difference between the average values ​​of UL1(i)' and UL2(i)' and the difference between the average values ​​of UR1(i)' and UR2(i)' are the horizontal drift of the interferometric images between the two samplings.

[0043] Based on the above calculations, the horizontal distance of image drift during the observation period is shown in Figure 3(b). Figures 3(c) and (d) show the wind field inversion results before and after image drift correction, respectively. In Figure 3(c), the vertical axis represents the wind speed (m / s) before correction, and in Figure 3(d), the vertical axis represents the wind speed (m / s) after correction. The horizontal axis of both figures is Universal Time (UT). The black dotted line Zenith represents the zenith wind speed, the green dotted line North represents the northward horizontal wind speed, the blue dotted line South represents the southward horizontal wind speed, the red dotted line East represents the eastward horizontal wind speed, and the orange dotted line West represents the westward horizontal wind speed. Without correction, the vertical wind inversion results show significant changes with the continuous change of ambient temperature, and there is a strong correlation between the two. At the same time, the horizontal wind also shows a significant directional drift. After correction, these characteristics no longer appear, indicating that the influence of ambient temperature has been reduced to a sufficiently low level.

[0044] The present invention also provides an image drift monitoring system for an asymmetric spatial heterodyne interferometer, comprising a processor and a memory, wherein the memory stores a computer program, and the computer program, when running in the processor, implements the steps of the above-mentioned specific method for monitoring image drift.

[0045] The present invention also provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the above-described method for monitoring image drift.

[0046] This invention proposes a novel monitoring method to address the image drift problem in asymmetric spatial heterodyne interferometers. This method utilizes the constant feature width of the same scale to extract and match scale pattern features, thereby enabling precise calculation of the scale position obtained from two consecutive samplings.

[0047] This invention eliminates the traditional method of fitting the scale signal function, thus reducing the stringent requirements on optical processing technology. In practical applications, a grating scale can be established using only a common commercial laser marking machine, reducing equipment costs and operational complexity.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring image drift in an asymmetric spatial heterodyne interferometer, characterized in that, include: The step of constructing the scale pattern is to construct a periodic scale pattern in the edge region of the grating of the asymmetric spatial heterodyne interferometer. The image acquisition step involves acquiring the first and second interferometric images formed by the asymmetric spatial heterodyne interferometer, respectively. In the signal selection and interpolation step, multiple rows of pixels with scale patterns are selected in the interferometric image and averaged to obtain one-dimensional interferometric image data containing the scale pattern. The horizontal axis represents the position of the pixel and the vertical axis represents the amplitude of the image. Linear interpolation is performed on one-dimensional interferometric image data to expand the data volume; The step of extracting scale pattern features involves calculating the feature amplitude of the scale pattern edge for the one-dimensional interferometric image data. The step of matching scale pattern features involves aligning the left and right feature edges of each scale pattern, selecting points with the same amplitude on the left and right edges of each scale pattern, and subtracting the horizontal coordinates of the left and right edges under the same amplitude to obtain the feature width W1 of the scale pattern in the first interference image and the feature width W2 of the scale pattern in the second interference image; comparing W1 and W2, if the difference between the two is less than a set value of pixels, then feature matching is complete. In the step of calculating the drift, the average value of the horizontal coordinates of the left and right edges of the scale pattern in the first and second interferometric images after feature matching is calculated respectively, and the difference between the two is the horizontal drift of the two sampled images.

2. The image drift monitoring method according to claim 1, characterized in that, The step of extracting scale pattern features further includes: selecting data points in the scale signal that are at the feature amplitude, expanding them on the left and right sides respectively, selecting more additional data points, and performing grouping operations based on the continuity of their position coordinates. The obtained amplitude and position relationship information is the extracted scale pattern feature.

3. The image drift monitoring method according to claim 2, characterized in that, The step of extracting scale pattern features also includes: verifying the accuracy of the expanded and grouped data, based on the monotonicity of the data and the positional interval of the average amplitude of each group; if a group of data is non-monotonic, or the positional interval exceeds a specific range of the scale period, the data grouping is determined to be incorrect; then the expansion distance is reduced, and the grouping operation is repeated until the grouping result passes the verification or the image is determined to be abnormal and unable to extract effective features.

4. The image drift monitoring method according to claim 1, characterized in that, In the step of extracting the scale pattern features, calculating the feature amplitude of the scale pattern edge includes, For a one-dimensional interferometric image signal, the peak and valley values ​​are extracted respectively, and the average value M of the peak value and the average value N of the valley value are calculated. (M+N) / 2 is used as the characteristic amplitude of the scale pattern edge.

5. The image drift monitoring method according to claim 1, characterized in that, Before the signal selection and interpolation step, an abnormal pixel correction step is also included, in which abnormal peak pixels with higher signal strength than surrounding pixels are identified by using a preset threshold to determine the first and second interference images, and the abnormal peak pixels are replaced by the average value of adjacent pixels in the same column.

6. The image drift monitoring method according to claim 5, characterized in that, The method for determining the preset threshold includes: dividing the image into several sub-blocks, calculating the independent mean amplitude and standard deviation within each sub-block, and setting the threshold for the i-th sub-block as follows: ,in It is the local pixel mean. It is the standard deviation, and k is a constant ranging from 3 to 5.

7. The image drift monitoring method according to claim 1, characterized in that, In the signal selection and interpolation step, the data volume is increased by 1000 times; in the matching scale pattern feature step, the set value is 1 / 1000.

8. The image drift monitoring method according to claim 1, characterized in that, In the step of matching scale pattern features, aligning the left and right feature edges of each scale pattern specifically includes: for the i-th scale pattern of the first interferometric image, generating a uniform amplitude axis based on a set step size, and then mapping the original discrete amplitudes GL1(i) and GR1(i) on both sides of the left and right edges of the scale pattern onto the uniform amplitude axis using a linear interpolation method, and finally outputting the aligned left and right edge data UL1(i) and UR1(i); for the i-th scale pattern of the second interferometric image, generating a uniform amplitude axis based on a set step size, and then mapping the original discrete amplitudes GL2(i) and GR2(i) on both sides of the left and right edges of the scale pattern onto the uniform amplitude axis using a linear interpolation method, and finally outputting the aligned left and right edge data UL2(i) and UR2(i). The step of matching scale pattern features further includes: calculating the feature width W1(i) of each scale pattern with the same amplitude based on UL1(i) and UR1(i); calculating the feature width W2(i) of each scale pattern with the same amplitude based on UL2(i) and UR2(i); comparing the feature widths contained in W1(i) with the feature widths contained in W2(i) one by one, and if there is a set of feature widths with a difference less than the set value of pixels, then the feature matching of the i-th scale is considered to be completed; the position coordinates of the first interference image scale pattern and the second interference image scale pattern that satisfy this condition are UL1(i)', UR1(i)' and UL2(i)', UR2(i)'; The step of calculating the drift amount specifically includes: calculating the average values ​​of UL1(i)' and UR1(i)' and the average values ​​of UL2(i)' and UR2(i)' of multiple scale patterns respectively. The difference between the average values ​​of UL1(i)' and UL2(i)' and the difference between the average values ​​of UR1(i)' and UR2(i)' are the horizontal drift amounts of the first and second interferometric images.

9. An image drift monitoring system for an asymmetric spatial heterodyne interferometer, comprising a processor and a memory, characterized in that, The memory stores a computer program, which, when run in the processor, implements the steps from image acquisition to drift calculation as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer program is stored therein, and when executed by a processor, the computer program implements the steps from the image acquisition step to the calculation of the drift amount as described in any one of claims 1 to 6.

Images