Interference fringe distortion correction method

Abstract

The invention relates to the field of vibration non-contact measurement, in particular to an interference fringe distortion correction method. The method comprises the steps that an interference fringe image is preprocessed, and interference of background light and high-frequency noise are eliminated; gradient amplitudes and gradient directions in the horizontal direction and the vertical direction are calculated respectively, and the edge of the interference fringes is determined through double-threshold detection; judging a distortion type according to the edge of the interference fringes; obtaining the center line slope of the interference fringes; searching a maximum response value through Fourier transform; determining accurate displacement through an extreme point of an interpolation function; different interference fringe distortions are quantized into offsets through a phase correlation function, and reverse compensation correction is carried out. Through interference fringe center line detection, the problem that interference fringe image distortion affects the detection precision of the coherent dispersion vibration meter is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vibration non-contact measurement, in particular to an interference fringe distortion correction method. BACKGROUND

[0002] As an important power source of nuclear power energy transmission, the nuclear power steam turbine is highly precise, and its health state is of great significance to the safe operation of the reactor. In the operation process, there are a large number of vibration abnormal phenomena of the nuclear power steam turbine due to process parameter fluctuations, equipment failures and external environmental interference. Therefore, in order to ensure the long-term safe operation of the generator, detect and locate the vibration abnormal data, and avoid the occurrence of destructive accidents, it is necessary to detect, locate and further analyze the vibration abnormal data.

[0003] The coherent dispersion vibration measurement method combining spectrum and interference is an interference measurement technology based on the interference principle of low coherence light, which has the advantages of low cost, high measurement precision and non-destructive, and can well meet the requirements of precise vibration measurement.

[0004] The coherent dispersion imaging device needs to provide higher quality interference fringe images, and accurately record the phase changes of the interference fringes for high-precision data inversion. After the target signal passes through the coherent dispersion imaging device, a two-dimensional interference fringe is formed on the detector, and the errors of each link such as the mirror system, the focal plane system, the spectrometer system and the detector will affect the form of the interference fringe to a certain extent, so that the two-dimensional interference fringe received by the detector is distorted in different types. These errors include some inherent defects of the coherent dispersion imaging device itself and the changes of the environment in which the device is located, and the interference fringe distortion formed thereby includes interference fringe tilt, interference fringe offset and interference fringe bending and other types. The interference fringe distortion will interfere with the phase information of the fringe, and will have a great influence on the subsequent data inversion. SUMMARY

[0005] The technical problem to be solved by the present application is to provide an interference fringe distortion correction method, which solves the problem of interference fringe image distortion affecting the detection precision of the coherent dispersion vibration measurement instrument by detecting the center line of the interference fringe.

[0006] The present application provides an interference fringe distortion correction method, comprising the following steps:

[0007] Step 1: pre-process the interference fringe image to eliminate the interference of background light and high-frequency noise;

[0008] Step 2: calculate the gradient amplitude and gradient direction in the horizontal and vertical directions respectively, and determine the edge of the interference fringe through double-threshold detection;

[0009] Step 3: judge the distortion type according to the edge of the interference fringe;

[0010] Step 4: Obtain the center line slope of the interference fringes;

[0011] Step 5: Find the maximum response value by Fourier transform;

[0012] Step 6: Determine the accurate displacement by the extreme point of the interpolation function;

[0013] Step 7: Quantize the distortion of different interference fringes into the offset by the phase correlation function, and perform reverse compensation correction.

[0014] In an embodiment of the present application, in step 1, the background light intensity is subtracted from the original light intensity I raw (x,y) to obtain the corrected light intensity distribution I corrected (x,y) by shooting a reference image without interference fringes.

[0015] The interference fringe image is weighted and averaged by using Gaussian filtering to suppress high-frequency noise.

[0016] In an embodiment of the present application, step 2 specifically includes:

[0017] Step 2-1: The Sobel operator is used to calculate the gradient of each pixel in the image in the horizontal and vertical directions respectively; according to the gradients in the two directions, the gradient amplitude and gradient direction of each pixel are calculated;

[0018] Step 2-2: Discretize the gradient direction into four main directions, compare the gradient amplitudes of the expected neighborhood pixels of each pixel in the gradient direction, and only keep the pixel with the maximum local gradient;

[0019] Step 2-3: Sort the gradient amplitudes of all pixels from high to low, and set high and low thresholds; the pixels higher than the high threshold are strong edges, and the pixels between the high and low thresholds are weak edges;

[0020] Step 2-4: Keep all strong edge pixels, and traverse all weak edge pixels; if a weak edge pixel is connected with a strong edge pixel in the neighborhood, it is promoted to a strong edge, otherwise it is discarded;

[0021] Step 2-5: Obtain the binarized and continuous upper and lower edges of the interference fringes.

[0022] In an embodiment of the present application, the high threshold is the gradient amplitude corresponding to the top 10% of the pixels, and the gradient amplitude corresponding to the pixels between 10% and 30%.

[0023] In an embodiment of the present application, step 4 specifically includes:

[0024] Step 4-1: Obtain the centroid of each row of the interference fringes inside the edge, perform spline fitting on the discrete centroid points by using spline interpolation, generate a smooth curve to obtain the center line of the interference fringes;

[0025] Step 4-2: Calculate the slope of the center line to obtain the tilt angle of the interference fringes, wherein the tilt angle of the interference fringes is:

[0026]

[0027] In one embodiment of the present application, in step 4-1, the centroid of each row is obtained by calculating the weighted average of the interference fringes inside the edge row by row.

[0028] In one embodiment of the present application, in step 4-1,

[0029] For the center line extracted row by row, the gradient in the horizontal direction of adjacent centroids is calculated, and the points with a gradient greater than 1.5 pixels are determined as abnormal points, and the neighborhood of the abnormal points is smoothed.

[0030] In one embodiment of the present application, step 5 specifically comprises:

[0031] Step 5-1: Select the wavelength channel at the center of the interference fringes as the reference column I1(x, y), and the remaining wavelength channels as the offset columns I2(x, y), and perform fast Fourier transform on I1(x, y) and I2(x, y) respectively;

[0032] Step 5-2: Calculate the normalized cross power spectrum;

[0033] Step 5-3: Perform inverse Fourier transform to obtain the phase correlation map;

[0034] Step 5-4: Find the maximum response value coordinates based on the phase correlation map.

[0035] In one embodiment of the present application, step 6 specifically comprises:

[0036] Centering on the position of the maximum response value, the response values of the position and its neighborhood are intercepted as sampling points;

[0037] An interpolation function is used to fit the sampling points;

[0038] The sub-pixel accurate displacement is obtained by solving the extreme condition.

[0039] In one embodiment of the present application, the interpolation function is:

[0040]

[0041] Wherein B(·) is Catmull-Rom spline:

[0042]

[0043] (x0,y0) is the position coordinate of the maximum response value.

[0044] Compared with the prior art, the interference fringe distortion correction method has the following beneficial effects:

[0045] (1) The centroid of the interference fringe is calculated to determine the tilt angle of the interference fringe, and the weighted average mechanism has natural inhibition ability to random noise. At the same time, the Gaussian filtering and double threshold processing can also significantly reduce the interference of noise. Therefore, the present application has strong robustness.

[0046] (2) Through non-maximum suppression and double threshold processing, the edge positioning is accurate and fine. The inverse Fourier transform is used to position the pulse peak, and the pulse peak value is interpolated to realize sub-pixel level displacement estimation. The detection accuracy is high.

[0047] (3) The present application calculates the centroid of the interference fringe by row / column to determine the tilt angle of the interference fringe; only the position difference of adjacent points is needed to eliminate the abnormal points of the interference fringe edge. The calculation complexity is low, and it is suitable for real-time processing. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 The flow chart of the interference fringe distortion correction method is shown. DETAILED DESCRIPTION

[0049] In order to further understand the present application, the embodiments of the present application are described below in conjunction with the examples, but it should be understood that these descriptions are only for further illustrating the features and advantages of the present application, and are not limitations of the present application.

[0050] The embodiments of the present application disclose an interference fringe distortion correction method, as shown in Figure 1 The method comprises the following steps:

[0051] Step 3: judging the distortion type according to the edge of the interference fringe;

[0052] Step 1: pre-processing the interference fringe image to eliminate the interference of background light and high-frequency noise;

[0053] By shooting a reference image without interference fringe, the background light intensity is subtracted from the light intensity I raw (x,y) of the original image to obtain the corrected light intensity distribution I corrected (x,y);

[0054] I corrected (x,y)=I raw (x,y)-I background(x, y)

[0055] I background (x, y) is the background light intensity;

[0056] The interference fringe image is weighted and averaged by using Gaussian filtering to suppress high-frequency noise while retaining the main edge structure.

[0057] Step 2: Calculate the gradient amplitude and gradient direction in the horizontal and vertical directions respectively, and determine the edges of the interference fringes by double-threshold detection;

[0058] Specifically, it includes:

[0059] Step 2-1: Use the Sobel operator to calculate the gradient of each pixel in the image in the horizontal and vertical directions respectively; according to the gradient in the two directions, calculate the gradient amplitude and gradient direction of each pixel;

[0060] The gradient amplitude represents the intensity of the edge of each pixel; the gradient direction represents the orientation of the edge;

[0061] Step 2-2: Discretize the gradient direction into four main directions, 0°, 45°, 90°, and 135°, respectively, compare the gradient amplitude of the expected neighborhood pixels of each pixel in the gradient direction, and only keep the pixel with the maximum local gradient;

[0062] This step eliminates redundant responses of the edge and improves the edge positioning accuracy;

[0063] Step 2-3: Sort the gradient amplitudes of all pixels from high to low, and set high and low thresholds; pixels higher than the high threshold are strong edges, and pixels between the high and low thresholds are weak edges;

[0064] The high threshold is the gradient amplitude corresponding to the top 10% of pixels, and the gradient amplitude corresponding to the pixels ranked between 10% and 30%.

[0065] Step 2-4: Keep all strong edge pixels, and traverse all weak edge pixels; if a weak edge pixel is connected to a strong edge pixel in the neighborhood, it is promoted to a strong edge, otherwise it is discarded;

[0066] Step 2-5: Get the binarized and continuous upper and lower edges of the interference fringes.

[0067] Step 3: Determine the distortion type according to the edges of the interference fringes;

[0068] When the interference fringes are offset along the dispersion direction, all the interference fringes are synchronously offset, and the final calculation result is not affected; under the spatial restriction of the slit, the interference fringes are widened as the interference fringe sequence of different wavelengths along the dispersion direction, so when the upper and lower boundaries of the interference fringes are not aligned, it indicates that part of the interference fringes is offset along the fringe direction; preferably, whether there is an angle point in the upper and lower boundaries of the interference fringes is detected, and if there is an angle point, it indicates that part of the interference fringes is offset along the fringe direction.

[0069] Step 4: obtaining the center line slope of the interference fringes;

[0070] Specifically, the method comprises the following steps:

[0071] Step 4-1: obtaining the centroid of each row of the interference fringes inside the edge, performing spline interpolation on the discrete centroid points to generate a smooth curve to obtain the center line of the interference fringes;

[0072] Specifically, since the interference fringes are perpendicular to the horizontal line, the centroid of each row is obtained by calculating the weighted average of the interference fringes inside the edge row by row;

[0073] The calculation formula is:

[0074]

[0075] Wherein: x c (y) is the centroid position of the yth row, and I(x, y) is the light intensity value at the coordinates (x, y).

[0076] The discrete centroid points are spline fitted by spline interpolation to generate a smooth curve to obtain the center line of the interference fringes,

[0077]

[0078] Wherein: x c (y+k) is the original center line coordinate of the y+kth row, N is the radius of the filtering window, is the center line coordinate of the yth row after smoothing;

[0079] For the center line extracted row by row, the gradient of the adjacent centroid in the horizontal direction is calculated:

[0080] Δx c (y) = x c (y+1) - x c (y)

[0081] x c (y) is the centroid position of the yth row, x c (y+1) is the centroid position of the y+1th row;

[0082] Δx c(y) Points greater than 1.5 pixels are determined as abnormal points, and the neighborhood of the abnormal points is smoothed to suppress mutations;

[0083] Step 4-2: Calculate the slope of the center line to obtain the tilt angle of the interference fringes;

[0084] The tilt angle is the angle of the interference fringes corresponding to the ideal position offset;

[0085] Let the coordinates of the points on the center line be (y i ,x i ), and there are n points. The slope k is the angle θ of the interference fringes corresponding to the ideal position offset:

[0086]

[0087] By calculating the slope of the center line of the interference fringes, the tilt angle of the interference fringes can be obtained, which finally manifests as different pixels having different sizes of offsets along the dispersion direction / interference fringe direction.

[0088] Step 5: Find the maximum response value by Fourier transform;

[0089] Specifically, it includes:

[0090] Step 5-1: Select the wavelength channel at the center of the interference fringes as the reference column I1(x,y), and the remaining wavelength channels as the offset columns I2(x,y). Perform fast Fourier transform on I1(x,y) and I2(x,y) respectively;

[0091] Step 5-2: Calculate the normalized cross-power spectrum;

[0092] Step 5-3: Perform inverse Fourier transform to obtain the phase correlation map;

[0093] Step 5-4: Find the maximum response value coordinates based on the phase correlation map;

[0094] Detect the pulse peak, and find the position coordinates (x0,y0) of the maximum response value in the phase correlation map.

[0095] Step 6: Determine the accurate displacement by the extreme point of the interpolation function;

[0096] Specifically, it includes:

[0097] Take the position of the maximum response value as the center, and intercept the response values of the position of the maximum response value and its neighborhood as sampling points;

[0098] Use an interpolation function to fit the sampling points;

[0099] Taking the position (x0, y0) of the maximum response value as the center, the response values of a 4x4 neighborhood are intercepted, the neighborhood data is fitted using a cubic polynomial, cubic interpolation is performed in the horizontal direction for each row to generate 4 interpolation curves, and then the 4 interpolation curves are again cubically interpolated in the vertical direction to construct a two-dimensional surface.

[0100] The interpolation function is:

[0101]

[0102] The estimated value of the phase correlation map at the sampling point is calculated using the interpolation function;

[0103] Where (x, y) is the target interpolation position, (x0, y0) is the position coordinate of the maximum response value, p(x0+i, y0+j) is the response value of the phase correlation map at the integer grid point, and B(·) is a cubic basis function:

[0104]

[0105] t is a normalized distance parameter.

[0106] The sub-pixel accurate displacement is obtained by solving the extreme value condition.

[0107] Step 7: Quantize the different interference fringe distortion into a shift amount by the phase correlation function, and perform reverse compensation correction.

[0108] The above description of the embodiments is only used to help understand the method of the present application and its core idea. It should be noted that, for those skilled in the art, without departing from the principles of the present application, some improvements and modifications can be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

[0109] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of correcting for interference fringe distortion, characterized by, The method comprises the following steps: Step 1: pre-process the interference fringe image to eliminate the interference of background light and high-frequency noise; Step 2: calculate the gradient amplitude and gradient direction in horizontal and vertical directions respectively, and determine the edge of the interference fringe through double-threshold detection; Step 3: determine the distortion type according to the edge of the interference fringe; Step 4: obtain the center line slope of the interference fringe; Step 5: find the maximum response value through Fourier transform; Step 6: determine the accurate displacement through the extreme point of the interpolation function; Step 7: quantize the distortion of different interference fringes into offset through a phase correlation function, and perform reverse compensation correction.

2. The interference fringe distortion correction method according to claim 1, characterized by, In step 1, the corrected intensity distribution I raw (x,y) is obtained by subtracting the background intensity from the original intensity I corrected (x,y) in the reference image taken without interference fringes. The interference fringe image is weighted and averaged through Gaussian filtering to suppress high-frequency noise.

3. The interference fringe distortion correction method according to claim 1, characterized by, The step 2 specifically comprises: Step 2-1: calculate the gradient of each pixel in the image in horizontal and vertical directions respectively using a Sobel operator; according to the gradient in the two directions, calculate the gradient amplitude and gradient direction of each pixel; Step 2-2: discretize the gradient direction into four main directions, compare the gradient amplitude of the expected neighborhood pixels of each pixel in the gradient direction, and only keep the pixel with the maximum local gradient; Step 2-3: sort the gradient amplitudes of all pixels from high to low, and set a high threshold and a low threshold; the pixels higher than the high threshold are strong edges, and the pixels between the high threshold and the low threshold are weak edges; Step 2-4: keep all strong edge pixels, and traverse all weak edge pixels; if a weak edge pixel is connected with a strong edge pixel in the neighborhood, it is promoted to a strong edge, otherwise it is discarded; Step 2-5: obtain the binarized and continuous upper and lower edges of the interference fringe.

4. The interference fringe distortion correction method according to claim 3, characterized by, The high threshold is the gradient amplitude corresponding to the top 10% of the pixels, and the gradient amplitude corresponding to the pixels between 10% and 30%.

5. The interference fringe distortion correction method according to claim 1, characterized by, The step 4 specifically comprises: Step 4-1: obtain the centroid of each row of the interference fringe inside the edge, perform spline fitting on the discrete centroid points to generate a smooth curve, and obtain the center line of the interference fringe; Step 4-2: calculate the slope of the center line to obtain the inclination angle of the interference fringe, wherein the inclination angle of the interference fringe is:

6. The interference fringe distortion correction method according to claim 5, characterized by, In the step 4-1, the centroid of each row is obtained by calculating the weighted average of the interference fringe inside the edge row by row.

7. The interference fringe distortion correction method according to claim 6, characterized by, In the step 4-1, For the center line extracted row by row, the gradient of adjacent centroids in the horizontal direction is calculated, and the point with a gradient greater than 1.5 pixels is determined as an abnormal point, and the neighborhood of the abnormal point is smoothed.

8. The interference fringe distortion correction method according to claim 1, characterized by, The step 5 specifically comprises: Step 5-1: select the wavelength channel at the center of the interference fringe as the reference column I1(x,y), and the remaining wavelength channels as the offset columns I2(x,y), and perform fast Fourier transform on I1(x,y) and I2(x,y) respectively; Step 5-2: calculate the normalized mutual power spectrum; Step 5-3: perform inverse Fourier transform to obtain a phase correlation graph; Step 5-4: find the maximum response value coordinate based on the phase correlation graph.

9. The interference fringe distortion correction method according to claim 1, characterized by, The step 6 specifically comprises: Take the position of the maximum response value as the center, and take the response value of the position and its neighborhood around the maximum response value as the sampling points; fit the sampling points using an interpolation function; By solving the extreme condition, the sub-pixel accurate displacement is obtained.

10. The interference fringe distortion correction method according to claim 9, characterized by, The interpolation function is: Where B(·) is Catmull-Rom spline: (x0, y0) is the position coordinate of the maximum response value.

Images