A measurement method for a quantum dot grating scale

Abstract

The present invention provides a measurement method for a quantum dot grating scale. The method includes four stages: image preprocessing, track detection, track skeleton extraction, and displacement measurement. First, preprocess the track image of the quantum dot grating scale, such as filtering and binarization. Then, after obtaining the discrete points of the upper and lower edges of the track, perform interpolation to obtain the track contour, thereby completing the detection of the track. According to the obtained upper and lower edges, obtain the track center line, and then perform triangular wave fitting to obtain the track skeleton. Finally, use the vertices of the triangular wave as the symbol element boundaries to measure, compensate, and denoise the displacement value. The method of the present invention calculates the fine displacement value from coarse to fine, can well perform fast and accurate measurement of the displacement value according to the image characteristics of the quantum dot grating scale, reduces the amount of data processing, and speeds up the algorithm operation speed. The obtained wave peak vertices, that is, the symbol element boundaries, have high stability, can achieve the effect of real-time suppressing measurement noise, and have the advantage of being resistant to noise influence.

Description

Technical Field

[0001] The present invention relates to the technical field of quantum dot grating scales, and more specifically, to a measurement method for quantum dot grating scales. Background Art

[0002] With the development of modern industry, ultra-precision manufacturing and processing in the micro-nano field have attracted increasing attention from the international academic and industrial communities. Modern micro-nano precision manufacturing equipment is a unified entity with global coupling of multiple physical processes, and is usually a complex system integrating functional units in multiple disciplinary fields such as machinery, electronics, hydraulics, and control. An accurate positioning feedback system is equivalent to the eyes of the equipment, improving the reliability and operability of the equipment and meeting all requirements such as the monitoring of the stroke and the measurement of the positioning accuracy during the equipment production and manufacturing process. The existing mainstream displacement sensors include laser interferometers and grating scales.

[0003] A laser interferometer needs to cooperate with various optical mirrors, etc. to measure linear position and speed. It mainly achieves sub-nanometer level positioning accuracy and communication resolution by counting and subdividing interference fringes. In practical applications, the use of a laser interferometer overly relies on the measurement environment, and at the same time, its large volume and many optical parts make it inconvenient to be integrated into the equipment, etc., which brings certain difficulties to practical production applications.

[0004] Compared with a laser interferometer, a grating scale has lower cost and higher actual production utilization rate, and has advantages such as low cost and high flexibility. It is the core measurement device for micro-nano ultra-precision processing equipment and is widely used in high-precision equipment such as high-end CNC machine tools, robots, tracking systems, and coordinate measuring machines. Grating scales are divided into absolute grating scales and incremental grating scales. Among them, the incremental grating scale can match a faster working speed and has a lower usage cost, but its measurement highly relies on the engraving of grating markers in the code track. The limitations of traditional grating manufacturing processes lead to insufficient grating size and accuracy, low yield due to accuracy loss, difficulty in adjusting the microscopic aspect ratio, etc., and high manufacturing cost and long cycle. In addition, grating scales based on grid line measurement all require a follow-up discrete light source system, which has an impact on the microenvironment temperature inside the grating measurement device, is difficult to compensate, and is also difficult to achieve nano-level measurement.

[0005] And such as Figure 1As shown in the figure, the electrospinning near-field direct writing technology that combines additive manufacturing with novel nano-level display quantum dot materials can fabricate a quantum dot grating scale, which can overcome the above-mentioned drawbacks. It changes the existing structure where the light source and the reading head of the grating scale coexist, and directly reads using the luminescent property of quantum dots, avoiding the use of the integrated structure of the reading head and the light source and the thermal interference phenomenon of the heat generated by the light source causing thermal deformation of the grating carrier. At the same time, it can be seen that there are significant differences between the patterns of the traditional grating scale track and the quantum dot grating scale track. Therefore, the existing traditional grating scale measurement methods cannot be applied to the quantum dot grating scale.

[0006] A patent for a decoding method of an incremental encoder is disclosed in the prior art. This patent collects the level states of the A-phase pulse and the B-phase pulse emitted by the incremental encoder; when both the A-phase pulse and the B-phase pulse change, according to the sequence of changes of the A-phase pulse and the B-phase pulse and the change types of the A-phase pulse and the B-phase pulse, the rotation direction of the incremental encoder is marked. This patent method can reduce the requirements for the signal stability of the incremental encoder, improve the anti-interference ability of decoding, reduce the probability of error codes and lost codes, and improve the decoding accuracy; however, there is little report on how to achieve fast and accurate measurement of the displacement value for the image characteristics of the quantum dot grating scale. Summary of the Invention

[0007] The present invention provides a measurement method for a quantum dot grating scale, which realizes the measurement of the quantum dot grating scale from rough to precise.

[0008] In order to achieve the above technical effects, the technical solution of the present invention is as follows:

[0009] A measurement method for a quantum dot grating scale includes the following steps:

[0010] S1: Perform filtering and binarization preprocessing on the quantum dot grating scale track image;

[0011] S2: After obtaining the discrete points of the upper and lower edges of the track, perform interpolation to obtain the track contour, thereby completing the detection of the track;

[0012] S3: Obtain the track center line based on the obtained upper and lower edges, and then perform triangular wave fitting to obtain the track skeleton;

[0013] S4: Use the vertex of the triangular wave as the code element boundary to perform displacement value measurement, compensation, and denoising.

[0014] Further, in the step S2, the specific process of detecting the track is:

[0015] Using the clear area of the image as the region of interest (ROI), every k pixels along the X-axis, where k = 15, find the first white pixel after binarization from top to bottom and from bottom to top at the upper and lower edges of the ROI respectively, as the discrete edge points of the code track , , after extracting the discrete points, perform cubic spline interpolation on the discrete points at the upper and lower edges. The process is as follows. First, calculate the step size , and the expression is as shown in Equation (1):

[0016]

[0017] Then substitute the edge discrete points and the step size into the matrix equation (2) and solve for :

[0018]

[0019] Substitute into Equations (3), (4), (5), and (6) to obtain the coefficients of the spline curve , , , :

[0020]

[0021]

[0022]

[0023]

[0024] In each sub-interval , create Equation (7) and substitute the values of the edge discrete points in the sub-interval into Equation (7) to complete the interpolation:

[0025]

[0026] Obtain the upper edge and the lower edge .

[0027] Furthermore, in step S3, the process of extracting the code track skeleton is as follows:

[0028] Obtain the code track center line from the upper edge and the lower edge , as shown in Equation (8):

[0029]

[0030] After performing a triangular wave fitting on the center of the code track using the least squares method, the code track skeleton is obtained. The specific implementation of the least squares fitting algorithm is as follows:

[0031] Obtain the code track center line data ( , ), assuming that they should satisfy the trigonometric function relationship , where the trigonometric function relationship is as shown in Equation (10). Based on this known information, the parameters of the function f need to be determined and . If represents the parameters to be determined in the function, then the goal is to find a set of such that the value of the following Equation (9), is minimized:

[0032]

[0033] .

[0034] Furthermore, in the step S4, the process of displacement measurement is as follows:

[0035] Using the obtained code track skeleton, through the code track displacement calculation method from coarse to fine, the accurate displacement value is calculated satisfying the expression (11), where , are the coarse and fine displacement positioning values respectively:

[0036] .

[0037] Furthermore, the calculation process of the coarse displacement positioning value is as follows:

[0038] After the ruler returns to zero, using the clear image area as the region of interest ROI, obtain the X-axis coordinate of the uppermost peak closest to the left boundary of the ROI of the code track skeleton through the method of extracting the code track skeleton described in claim 3 The column where it is located is used as the fixed observation position; when the ruler undergoes displacement, the phase of the fitted skeleton waveform will change accordingly, so the value corresponding to the position in the picture will change up and down with the displacement of the ruler;

[0039] During the measurement process, set two thresholds up and down and . Only when the value in the observation position ( ) is first less than , and then greater than , is it considered that the code track displacement has passed through a cycle;

[0040] When the code track is displaced, each half symbol passes through the observation position once, and the counter accumulates correspondingly once as the rough measurement value , satisfies Equation (12), where is the symbol length:

[0041] .

[0042] Furthermore, the calculation process of the displacement fine positioning value is as follows:

[0043] The sub - divided measurement value is determined according to the distance from the observation point to the nearest wave peak, where is the physical length of one symbol of the code track, satisfies Equation (13), where the x - axis coordinate of the wave peak closest to the current distance from the observation position ( ), is the physical length corresponding to each pixel:

[0044] .

[0045] Furthermore, in step S4, after obtaining the displacement value, the data is input into the radial basis function neural network RBFNN to obtain a more accurate displacement value after non - linear compensation. Finally, the displacement value after non - linear compensation is input into the Kalman filter to filter the displacement value and suppress the measurement noise in real - time.

[0046] Furthermore, in step S4, the process of using the radial basis function neural network to perform non - linear compensation on the displacement value is as follows:

[0047] Measure the same code track multiple times to obtain n groups of measurement data before compensation , where is the true displacement value of the code track, as shown in Equation (14). Calculate the average value of the n groups of data to obtain the measurement average data :

[0048]

[0049] Take ( ) as the training sample and the true value and input them into the radial basis function neural network, where φ(x) is the radial basis function and satisfies Equation (15). In subsequent measurements, substitute the displacement value into the trained radial basis function neural network to achieve non - linear compensation:

[0050] .

[0051] Further, in step S4, the non-linear compensation and denoising of the displacement measurement value further includes using a Kalman filter to denoise the measured displacement value:

[0052] After obtaining a more accurate displacement value after non-linear compensation, the non-linearly compensated displacement value is input into the Kalman filter to filter the displacement value and suppress measurement noise in real time. The prediction formulas of the Kalman filter are shown in equations (16) and (17), and the update formulas are shown in (18), (19), and (20):

[0053]

[0054]

[0055]

[0056]

[0057] 。

[0058] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0059] The method of the present invention includes four stages: image preprocessing, code track detection, code track skeleton extraction, and displacement measurement. First, preprocessing such as filtering and binarization is performed on the quantum dot grating size track image. Then, discrete points on the upper and lower edges of the code track are obtained and interpolated to obtain the code track contour, thereby completing the detection of the code track. The center line of the code track is obtained based on the obtained upper and lower edges, and then triangular wave fitting is performed to obtain the code track skeleton. Finally, the vertices of the triangular wave are used as the code element boundaries to measure, compensate, and denoise the displacement value. The method of the present invention calculates a fine displacement value from coarse to fine, can well measure the displacement value quickly and accurately according to the image characteristics of the quantum dot grating scale, reduces the amount of data processing, and speeds up the algorithm operation speed. The obtained wave peak vertices, that is, the code element boundaries, have high stability, achieve the effect of real-time suppressing measurement noise, and have the advantage of resisting noise influence. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a comparison diagram of a quantum dot grating scale and a traditional incremental grating scale. Among them, Figure 1 in (a) is the morphology of the code track position mark of the quantum dot grating scale under an optical microscope, Figure 1 in (b) is the code track diagram of the traditional incremental grating scale;

[0061] Figure 2 is a flowchart of the method of the present invention;

[0062] Figure 3 is a process diagram of code track image preprocessing;

[0063] Figure 4 It is a schematic diagram for code track detection;

[0064] Figure 5 It is a schematic diagram for extracting the code track skeleton based on triangular wave fitting;

[0065] Figure 6 It is a schematic diagram for displacement value measurement;

[0066] Figure 7 It is a schematic diagram for the structure of a radial basis neural network. Specific implementation manners

[0067] The accompanying drawings are only for illustrative purposes and should not be construed as a limitation to this patent;

[0068] To better illustrate this embodiment, some components in the accompanying drawings will be omitted, enlarged or reduced, which do not represent the dimensions of the actual product;

[0069] For those skilled in the art, it is understandable that some well-known structures and their descriptions in the accompanying drawings may be omitted.

[0070] The technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0071] Embodiment 1

[0072] As Figure 2 shown, a measurement method for a quantum dot grating scale includes the following steps:

[0073] S1: Perform filtering and binarization preprocessing on the code track image of the quantum dot grating scale;

[0074] S2: After obtaining the discrete points of the upper and lower edges of the code track, perform interpolation to obtain the code track contour, thereby completing the detection of the code track;

[0075] S3: Obtain the code track center line according to the obtained upper and lower edges, and then perform triangular wave fitting to obtain the code track skeleton;

[0076] S4: Use the vertices of the triangular wave as the code element boundaries to perform displacement value measurement, compensation, and denoising.

[0077] Preprocessing of the code track image:

[0078] Since the image acquisition environment is extremely dark, there will be more noise points in the collected image data. Therefore, it is necessary to perform Gaussian filtering on the image to remove the noise points in the picture. Secondly, since the collected image is a color image, but the main emission color of the quantum dot grating scale is green, the green channel of the image is extracted and then binarized to extract the main information of the image and reduce the amount of image data processing, as Figure 3 shown.

[0079] Code track detection:

[0080] To improve the running efficiency of the algorithm, this method first obtains the discrete points of the upper and lower edges, and then performs cubic spline interpolation on the discrete points of the upper and lower edges respectively to obtain the upper and lower edges of the track, thereby obtaining the track profile.

[0081] Track skeleton extraction:

[0082] Traditional incremental grating rulers use the light and dark boundaries of the grating as the symbol boundaries, while the near-field direct writing printed serpentine track quantum dot grating ruler uses the peak of each symbol as the symbol boundary. Therefore, in order to obtain the fine peak coordinates, it is necessary to extract the skeleton of the track. The center line of the track is obtained through the upper and lower edges of the track, and then the center of the track is fitted with a triangular wave to obtain the skeleton, thereby obtaining the peak vertex coordinates of the waveform. The peak vertices obtained by this method have the advantages of high stability and resistance to noise.

[0083] Displacement measurement:

[0084] Coarse positioning of displacement value:

[0085] After the ruler returns to zero, the clear image area is used as the ROI (region of interest). The column where the x coordinate of the upper peak closest to the left boundary of the ROI is obtained by the above method as the fixed observation position. When the track is displaced, each time a symbol passes through the observation position, the counter accumulates the symbol length once as the rough measurement value.

[0086] Fine positioning of displacement value:

[0087] The fine positioning of the displacement value is obtained according to the distance between the observation position and the closest peak. The combination of coarse positioning and fine positioning gives the final displacement measurement value.

[0088] Nonlinear compensation and real-time denoising of displacement value

[0089] After obtaining the displacement value, the data is input into the RBFNN (radial basis function neural network) to obtain a more accurate displacement value after nonlinear compensation. Finally, the displacement value after nonlinear compensation is input into the Kalman filter to filter the displacement value and suppress the measurement noise in real time.

[0090] Embodiment 2

[0091] As Figure 2 shown, a measurement method for a quantum dot grating ruler includes the following steps:

[0092] S1: Filter and binarize the preprocessing of the quantum dot grating ruler track image;

[0093] S2: After obtaining the discrete points of the upper and lower edges of the track, interpolation is performed to obtain the track profile, thereby completing the detection of the track;

[0094] S3: Obtain the center line of the code track based on the obtained upper and lower edges, and then perform triangular wave fitting to obtain the code track skeleton;

[0095] S4: Use the vertices of the triangular wave as the code element boundaries to measure, compensate, and denoise the displacement values.

[0096] In step S2, the specific process of detecting the code track is as follows:

[0097] Take the clear area of the image as the region of interest ROI. Along the X-axis, every k pixels, find the first pixel point that is white after binarization from top to bottom and from bottom to top on the upper and lower edges of the ROI respectively, as the edge discrete points of the code track , . After extracting the discrete points, perform cubic spline interpolation on the upper and lower edge discrete points. The process is as follows: first calculate the step size , and the expression is as shown in Equation (1):

[0098]

[0099] Then substitute the edge discrete points and the step size into the matrix equation (2) to solve for :

[0100]

[0101] Substitute into Equations (3), (4), (5), and (6) to obtain the coefficients of the spline curve , , , :

[0102]

[0103]

[0104]

[0105]

[0106] In each sub-interval , create Equation (7), and substitute the values of the edge discrete points in the sub-interval into Equation (7) to complete the interpolation:

[0107]

[0108] Obtain the upper edge and the lower edge .

[0109] In step S3, the process of extracting the code track skeleton is as follows:

[0110] Through the upper edge and the lower edge the center line of the code track is obtained , as shown in Equation (8):

[0111]

[0112] Then, the triangular wave fitting of the code track center is performed by the least square method to obtain the code track skeleton. The specific implementation of the least square fitting algorithm is as follows:

[0113] Obtain the data of the center line of the code track ( , ), assuming that they should satisfy the trigonometric function relationship , where the trigonometric function relationship is as shown in Equation (10). Based on this known information, the parameters and of the function f need to be determined. If is used to represent the parameters to be determined in the function, then the goal is to find a set of to minimize the value of the following Equation (9), :

[0114]

[0115] .

[0116] In step S4, the process of displacement measurement is as follows:

[0117] Using the obtained code track skeleton, through the code track displacement calculation method from coarse to fine, the accurate displacement value is calculated satisfies Expression (11), where , are the coarse and fine displacement positioning values respectively:

[0118] .

[0119] The calculation process of the coarse displacement positioning value is as follows:

[0120] After the ruler returns to zero, using the clear image area as the region of interest ROI, the X-axis coordinate of the uppermost peak closest to the left boundary of the ROI obtained by the method of extracting the code track skeleton described in claim 3 The column where it is located is used as the fixed observation position; when the ruler moves, the phase of the fitted skeleton waveform will change accordingly, so in the picture The value corresponding to the position will change up and down with the displacement of the ruler;

[0121] Two thresholds, upper and lower, are set during the measurement and , only when the value in the observation position ( ) first is less than , and then is greater than , , is it considered that the code track displacement has passed through a cycle;

[0122] When the code track is displaced, the counter is incremented correspondingly once for each half symbol passing through the observation position, as the rough measurement value , satisfies Equation (12), where is the symbol length:

[0123] .

[0124] The calculation process of the fine positioning value of the displacement is:

[0125] The fine measurement value is determined according to the distance from the observation point to the nearest wave peak, where is the physical length of one symbol of the code track, satisfies Equation (13), where is the x-axis coordinate of the wave peak closest to the current observation position ( ), is the physical length corresponding to each pixel:

[0126] .

[0127] In step S4, after obtaining the displacement value, the data is input into the radial basis function neural network RBFNN to obtain a more accurate displacement value after non-linear compensation. Finally, the displacement value after non-linear compensation is input into the Kalman filter to filter the displacement value and suppress the measurement noise in real time.

[0128] In step S4, the process of using the radial basis function neural network to perform non-linear compensation on the displacement value is:

[0129] Perform multiple measurements on the same code track to obtain n groups of measurement data before compensation , where is the true displacement value of the code track, as shown in Equation (14). Calculate the average value of the n groups of data to obtain the measurement average data :

[0130]

[0131] Take ( ) as the training sample and the true value and input them into the radial basis function neural network, where φ(x) is the radial basis function and satisfies Equation (15). In subsequent measurements, the displacement value Substituting the trained radial basis neural network, nonlinear compensation can be achieved:

[0132] .

[0133] In step S4, performing nonlinear compensation and denoising on the displacement measurement value also includes using a Kalman filter to denoise the measured displacement value:

[0134] After obtaining a more accurate displacement value after nonlinear compensation, the displacement value after nonlinear compensation is input into the Kalman filter to filter the displacement value and suppress the measurement noise in real time. The prediction formula of the Kalman filter is shown in equations (16) and (17), and the update formula is shown in equations (18) and (19) and (20):

[0135]

[0136]

[0137]

[0138]

[0139] .

[0140] Example 3

[0141] like Figure 2 As shown, a measurement method for a quantum dot grating ruler includes the following steps:

[0142] 1. Code channel image preprocessing

[0143] First extract the green channel of the input color image, then perform binarization, and finally perform Gaussian filtering. Figure 3 shown.

[0144] 2. Code channel detection

[0145] Take the clear area of ​​the image as the region of interest ROI, and search for the first white pixel after binarization from the top and bottom edges of the ROI every k pixels along the X axis, as the edge discrete point of the code channel [ , ],like Figure 4 As shown in (a), after the discrete points are extracted, the upper and lower edge discrete points are interpolated by cubic spline. The process is as follows: first calculate the step length , as shown in formula (1):

[0146]

[0147] Then the edge discrete points and step size Substitute into the matrix equation (2) and solve to obtain :

[0148]

[0149] Substitute into equations (3), (4), (5), and (6) to obtain the coefficients of the spline curve , , , :

[0150]

[0151]

[0152]

[0153]

[0154] In each sub - interval , create equation (7) and substitute the values in the sub - interval of the edge discrete points into equation (7) to complete interpolation:

[0155]

[0156] The result is as shown in (b) of Figure 4 to obtain the upper edge and the lower edge .

[0157] 3. Extraction of the code - track skeleton

[0158] As shown in equation (8), obtain the code - track center line from the upper edge and the lower edge :

[0159]

[0160] The effect is as shown in (b) of Figure 4 . Then, perform triangular - wave fitting on the code - track center using the least - squares method to obtain the code - track skeleton, and the effect is as shown in Figure 5 .

[0161] Obtain the code - track center - line data ( , ). Assume that they should satisfy the trigonometric function relationship , where the trigonometric function relationship is as shown in equation (10). Based on this known information, the parameters and of the function f need to be determined. If we use Denote the parameters to be determined in the function. Then the goal is to find a set of such that for the following formula (9), the value is minimized:

[0162]

[0163] .

[0164] 4. Displacement measurement

[0165] Using the obtained code track skeleton above, through the code track displacement calculation method from coarse to fine, the displacement value is accurately calculated satisfying expression (11), where , are the coarse and fine displacement positioning values respectively:

[0166] .

[0167] 4.1 Coarse positioning of the code track displacement value

[0168] After the ruler returns to zero, take the clear image area as the ROI (region of interest). Through the above method, obtain the X-axis coordinate X of the uppermost peak closest to the left boundary of the ROI of the code track skeleton 0 The column where it is located is used as the fixed observation position, as shown by Figure 5 the white dashed line. When the ruler undergoes displacement, as shown by Figure 6 , the phase of the fitted skeleton waveform will change accordingly. Therefore, the y value corresponding to the position of X 0 in the picture will change up and down with the displacement of the ruler.

[0169] To improve the anti-interference ability of decoding, reduce the probability of error codes and lost codes, and improve the decoding accuracy. During the measurement process, set two thresholds Y u and Y d as shown by Figure 5 . Only when the y value in the observation position (X 0 , y) is first less than Y d , and then greater than Y u , is it considered that the code track displacement has passed a cycle.

[0170] When the code track is displaced, each half code element passes through the observation position once, and the counter is incremented accordingly as the coarse measurement value . Satisfying formula (12), where is the code element length.

[0171] .

[0172] 4.2 Fine positioning of the code track displacement value

[0173] Subdivided measurement value It is determined according to the distance from the observation point to the nearest wave crest. Among them is the physical length of one code element of the code track. Satisfies Equation (13), where The x-axis coordinate of the wave crest closest to the current distance observation position (X 0 , y), is the physical length corresponding to each pixel:

[0174]

[0175] 4.3 Nonlinear compensation and denoising of displacement measurement values

[0176] 4.3.1 Nonlinear compensation of displacement values using a radial basis neural network

[0177] Perform multiple measurements on the same code track to obtain n sets of measurement data before compensation , where is the true displacement value of the code track. As shown in Equation (14), calculate the average value of the n sets of data to obtain the measurement average data .

[0178]

[0179] Take ( ) as the training sample and the true value and input them into the radial basis neural network as Figure 7 shown, where φ(x) is the radial basis function and satisfies Equation (15). In subsequent measurements, substitute the displacement value into the trained radial basis neural network to achieve nonlinear compensation:

[0180]

[0181] 4.3.2 Denoising of measured displacement values using a Kalman filter

[0182] After obtaining a more accurate displacement value after nonlinear compensation, input the displacement value after nonlinear compensation into the Kalman filter to filter the displacement value and suppress measurement noise in real time. The prediction formulas of the Kalman filter are shown in Equations (16) and (17), and the update formulas are shown in (18), (19), and (20).

[0183]

[0184]

[0185]

[0186]

[0187]

[0188] Like or similar reference numerals correspond to like or similar components;

[0189] The positional relationships described in the drawings are for illustrative purposes only and should not be construed as limiting the present patent;

[0190] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A measurement method for a quantum dot grating scale, characterized in that, it includes the following steps: S1: Filter and binarize the preprocessing of the quantum dot grating scale track image; S2: After obtaining the discrete points of the upper and lower edges of the track, perform interpolation to obtain the track contour, thereby completing the detection of the track; S3: Obtain the track center line according to the obtained upper and lower edges, and then perform triangular wave fitting to obtain the track skeleton; S4: Use the vertex of the triangular wave as the code element boundary to measure, compensate, and denoise the displacement value; In the step S4, the process of displacement measurement is: Using the obtained code track skeleton, through a coarse-to-fine code track displacement calculation method, the displacement value is accurately calculated Satisfy expression (1), where and are the coarse and fine positioning values of the displacement respectively: The calculation process of the fine displacement positioning value is: Subdivided measurement value It is determined according to the distance from the observation point to the nearest wave crest, where is the physical length of one code element of the code track, satisfies Equation (2), where The x-axis coordinate of the wave crest closest to the current distance observation position ( ), is the physical length corresponding to each pixel: 。 2. The measurement method for a quantum dot grating scale according to claim 1, characterized in that, in the step S2, the specific process of detecting the track is: Taking the clear region of the image as the region of interest ROI, every k pixels along the X-axis, respectively find the first pixel that is white after binarization from the upper and lower edges of the ROI from top to bottom and from bottom to top, as the edge discrete points of the code track , , after extracting the discrete points, perform cubic spline interpolation on the upper and lower edge discrete points. The process is as follows. First, calculate the step size , and the expression is as shown in Equation (3): Then substitute the edge discrete points and the step size into the matrix equation (4) and solve to obtain : Substitute into equations (5), (6), (7), and (8) to obtain the coefficients of the spline curve , , , : In each sub-interval create Equation (9), and substitute the values in the sub-interval of the edge discrete points into Equation (9) to complete interpolation: Obtain the upper edge and the lower edge .

3. The measurement method for a quantum dot grating scale according to claim 2, characterized in that, in the step S3, the process of extracting the track skeleton is: Through the upper edge and the lower edge the center line of the code track is obtained , as shown in Equation (10): Then perform triangular wave fitting on the track center by the least squares method to obtain the track skeleton. The specific implementation of the least squares fitting algorithm is as follows: Obtain the center line data of the code track ( , ), assuming that they should satisfy the trigonometric function relationship , where the trigonometric function relationship is as shown in Equation (12). Based on this known information, the parameters of the function f and need to be determined. If is used to represent the parameters to be determined in the function, then the goal is to find a set of to minimize the value of the following Equation (11), : 。 4. The measurement method for a quantum dot grating scale according to claim 3, characterized in that, The calculation process of the coarse displacement positioning value is: After the ruler returns to zero, use the clear image area as the region of interest (ROI), and extract the X-axis coordinate of the upper peak closest to the left boundary of the ROI from the code track skeleton through the code track skeleton extraction. The column where it is located is used as the fixed observation position; when the ruler moves, the phase of the fitted skeleton waveform will change accordingly. Therefore, in the picture The corresponding value will change up and down with the displacement of the ruler. Set upper and lower thresholds during the measurement process and , and only when the value in the observation bit ( ) is first less than , and then greater than , is it considered that the code track displacement has passed through a cycle; When the code path is displaced, each half symbol passes through the observation position once, and the counter accumulates correspondingly once as the rough measurement value , satisfies Equation (13), where is the symbol length: 。 5. The measurement method for a quantum dot grating scale according to claim 4, characterized in that, In step S4, after obtaining the displacement value, input the data into the radial basis neural network RBFNN to obtain a more accurate displacement value after non-linear compensation. Finally, input the non-linearly compensated displacement value into the Kalman filter to filter the displacement value and suppress the measurement noise in real time.

6. The measurement method for a quantum dot grating scale according to any one of claims 2-5, characterized in that, k takes the value of 15.

Images