Three-dimensional data acquisition method based on galvanometer structured light

Abstract

The invention discloses a galvanometer structured light-based three-dimensional data acquisition method. The method comprises the following steps of 1, carrying out galvanometer structured light phase unwrapping; step 1.1, projecting a group of sinusoidal gratings to a measured object through a micro galvanometer laser generator; step 1.2, calculating a wrapped phase by using a four-step phase shift method; 1.3, unwrapping the wrapped phase by using a multi-frequency heterodyne method; step 2, grating phase filtering based on regularization; step 2.1, searching a smoothing function on a defined threshold L; 2.2, solving a minimum value problem of an energy function; the method has the beneficial effects that errors caused by defects of the laser are remarkably reduced or eliminated, the precision of three-dimensional reconstruction is improved, the robustness of the whole system is enhanced, the stable performance can be kept under various different measurement conditions, the precision and reliability of three-dimensional reconstruction are improved, and the method is suitable for large-scale popularization and application. And powerful technical support is provided for the development of the industrial three-dimensional measurement field.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial three-dimensional measurement technology, and in particular to a galvanometer structured light three-dimensional data acquisition method. BACKGROUND

[0002] In the field of three-dimensional reconstruction technology, the method based on galvanometer structured light has become an important technical means. This method mainly relies on a binocular matching technology based on grating phase, and its core is to use a galvanometer to reflect laser light and then generate sinusoidal fringe structured light with a specific pattern. This structured light is then accurately projected onto the surface of the object to be measured. By using a binocular camera system, multiple sinusoidal fringe images can be captured from different angles. After a series of processing steps, including decoding and phase unwrapping, a detailed phase data set can be obtained. This data set is indispensable in the three-dimensional reconstruction process, as it contains the three-dimensional information of the measured object surface.

[0003] However, in practical applications, due to the inherent limitations of laser hardware, the projected laser fringe often accompanies certain errors and noise. These errors and noise not only come from the physical properties of the laser, but also can be affected by environmental factors. These unfavorable factors have a significant negative impact on the subsequent phase unwrapping process, thereby reducing the accuracy and reliability of three-dimensional reconstruction. SUMMARY

[0004] To solve the above problems, the present application proposes an innovative solution, namely adopting a method based on second-order regular filter. Second-order regular filter algorithm is a commonly used filtering technique in signal processing, mainly used to remove noise and interference in the signal while retaining useful signal components. Second-order regular filter usually refers to second-order low-pass filter, and its core idea is to design a transfer function so that signals below a certain cutoff frequency can pass without attenuation, while signals above that frequency are attenuated. The purpose of this method is to effectively filter the laser fringe projected onto the object surface.

[0005] The technical solution adopted by the present application to achieve the above purpose is as follows: Step 1: Galvanometer structured light phase unwrapping method; Step 1.1: A set of sinusoidal gratings is projected onto the measured object by a micro-galvanometer laser generator, and the light intensity function modulated by the height of the measured object can be represented as: (1) In the formula, is the light intensity function, is represented as light intensity, for the object depth information, for the phase value of the image grating, is the phase of this object point.

[0006] Step 1.2: Calculate the wrapped phase using the four-step phase shift method; The four-step phase shift values can be set as:

[0007] From equation (1): (2) Solving the above four equations can obtain the phase function, that is: (3) Step 1.3: Use the method of multi-frequency heterodyne to unwrap the wrapped phase; Let the wrapped phase map of each set of stripes be: , and the wrapped phase map is obtained by formula (3) using three different frequencies, each frequency forms a geometric progression, the first frequency is 1, which can be set as , and then unwrap the wrapped phase maps of the three frequencies. Let the unwrapped phase map be , then the unwrapped formula of the three frequencies is: (4) Where subscript i represents the i-th set of projected grating images, and , represents the rounding operation.

[0008] Step 2: Regularization-based grating phase filtering; Step 2.1: The phase map generated by the galvanometer structured light according to formula (1) has certain noise, let be the observed image, and find a smooth function on the threshold value L , that is, the following process: (5) Where is the noise, and S is a subset of L.

[0009] Step 2.2: The process of regularization filtering can be regarded as the process of searching for the optimal solution of the inverse problem, that is, to establish a balance in the approximation degree between the obtained smooth filtered field and the observed data , that is, to solve the minimum value problem of the following energy function: (6) In formula (6), the first term is the approximation degree of the measured value directly with the observed value in the least square, and the second and third terms are the reciprocal of the continuous function space, that is, the deviation degree of the smoothness, characterization parameter of the model, the greater the value, the smoother the filter field.

[0010] Step 2.3: According to formula (6), the integral of the function is changed into the cumulative sum of the entire area, that is: (7) Where S is a subset of the sensing data set L.

[0011] Step 2.4: The image discrete expression of the regularization operator is as follows: (8) Step 2.5: According to formula (8), the discrete mathematical expression of the second-order regularization operator of the image can be derived as: (9) Step 2.6: According to formula (9), in order to obtain the gradient descent method based on iteration optimization can be used: that is (10) Where k is the iteration number.

[0012] The beneficial effects of the present application can significantly reduce or eliminate errors caused by defects of the laser itself. The second-order regularization filtering technology optimizes the stripe image through a mathematical model, thereby improving the image quality and ensuring the accuracy of the phase unwrapping process. The application of this technology not only improves the accuracy of three-dimensional reconstruction, but also enhances the robustness of the entire system, so that it can maintain stable performance under various measurement conditions.

[0013] The application of this filtering method will further improve the accuracy and reliability of three-dimensional reconstruction, providing strong technical support for the development of the industrial three-dimensional measurement field. DETAILED DESCRIPTION

[0014] The present application is a method for acquiring three-dimensional data based on a galvanometer structured light, which comprises the following steps: Step 1: Galvanometer structured light phase unwrapping method; Step 1.1: A group of sinusoidal gratings is projected onto the measured object by a micro-galvanometer laser generator, and the light intensity function modulated by the height of the measured object can be represented as: (1) In the formula, is the light intensity function, is the light intensity, is the object depth information,​ The phase value of the image grating. It is the phase of this object point.

[0015] Step 1.2: Calculate the wrapping phase using the four-step phase shift method; The four-step phase shift value can be set as follows:

[0016] From equation (1), we get: (2) Solving the above four equations simultaneously yields the phase function: (3) Step 1.3: Unwrap the wrapped phase using a multi-frequency heterodyne method; After the wrapped phase calculation is completed, the wrapped phase is unfolded using the multi-frequency heterodyne method. Let the wrapped phase diagram for each set of fringes be: The phase diagram is obtained by using three different frequencies through formula (3). Each frequency forms a geometric sequence, with the first frequency being 1, which can be set as 1. Then, the phase diagrams of the three frequencies are unfolded. Let the unfolded phase diagram be... Then the expansion formula for the three frequencies is: (4) Where the subscript i represents the i-th set of raster images projected, and , This indicates the rounding operation.

[0017] Step 2: Regularized grating phase filtering method; Step 2.1: The phase diagram generated by the galvanometer structured light according to formula (1) has a certain noise. Let... To observe the image, find a smoothing function defined on a threshold L. That is, the process is as follows: (5) in Let S be noise and S be a subset of L.

[0018] Step 2.2: The regularization filtering process can be viewed as a process of searching for the optimal solution to the inverse problem, that is, in the obtained smoothed filter field... With observation data To establish a balance based on the degree of approximation, we need to solve the minimum problem of the following energy function: (6) In formula (6), the first term represents the degree of approximation between the measured and observed values ​​under least squares conditions, while the second and third terms are... a smoothness degree of deviation, a representation parameter of the model, the larger the value, the smoother the filter field.

[0019] Step 2.3: According to formula (6), the integral of the product of the function and is changed into the cumulative sum of the entire area, that is: (7) where S is a subset of the sensing data set L.

[0020] Step 2.4: The image discrete expression of the intermediate regular operator is as follows: (8) Step 2.5: According to formula (8), the discrete mathematical expression of the second-order regular operator of the image can be derived as: (9) Step 2.6: According to formula (9), in order to obtain the gradient descent method based on iteration optimization can be used: that is (10) where k is the iteration number.

[0021] The application designs a grating phase filtering algorithm based on regularization. The core of the algorithm is to filter the observed phase image by using the second-order regularization operator. In the filtering process, the algorithm finds the optimal smoothing filter field by minimizing an energy function, which considers both the approximation degree between the filter field and the observed data and the smoothing degree of the filter field. In this way, the algorithm can effectively remove the noise and interference in the phase image while retaining useful phase information. In specific implementation, the algorithm first defines a smoothing function according to the observed phase image and a preset threshold. Then, the algorithm finds the optimal smoothing filter field by solving the minimum value problem of an energy function containing a least square term and a smoothing term. In this process, the algorithm uses an iterative optimization strategy based on the gradient descent method to continuously update the value of the smoothing filter field until the convergence condition is reached. Finally, the phase image processed by the grating phase filtering algorithm based on regularization will have higher signal-to-noise ratio and better smoothness, so as to more accurately reflect the three-dimensional information of the measured object surface.

Claims

1. A method for acquiring three-dimensional data based on galvanometer structured light, characterized in that: Includes the following steps: Step 1: Phase unfolding of structured light using galvanometers; Step 1.1: A set of sinusoidal gratings is projected onto the object under test using a micro-mirror laser generator. The light intensity function generated by the height modulation of the object under test is expressed as: (1) In the formula It is a function of light intensity. Expressed as light intensity, This is the object's depth information. The phase value of the image grating. It is the phase of this object point; Step 1.2: Calculate the wrapping phase using the four-step phase shift method; Step 1.3: Unwrap the wrapped phase using a multi-frequency heterodyne method; Step 2: Regularized grating phase filtering; Step 2.1: The phase diagram generated by the galvanometer structured light according to formula (1) has a certain noise. Let... To observe the image, find a smoothing function defined on a threshold L. ; Step 2.2: The regularization filtering process is viewed as a process of searching for the optimal solution to the inverse problem, that is, in the obtained smoothed filter field With observation data To establish a balance based on the degree of approximation, we need to solve the minimum problem of the following energy function: (6) In formula (6), the first term represents the degree of approximation between the measured and observed values ​​under least squares conditions, while the second and third terms are... An inverse of the first element in a continuous function space, representing the degree of deviation from smoothness. This represents the model's characterization parameters; the larger the value, the smoother the filtered field.

2. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 1, characterized in that: Step 1.2 specifically includes the following steps: The four-step phase shift value is set as follows: , From equation (1), we get: (2) Solving the above four equations simultaneously yields the phase function: (3)。 3. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 2, characterized in that: Step 1.3 specifically includes the following steps: Let the wrapping phase diagram of each set of stripes be: The phase diagram is obtained by using three different frequencies through formula (3). Each frequency forms a geometric sequence, with the first frequency being 1, which can be set as 1. Then, the phase diagrams of the three frequencies are expanded; let the expanded phase diagrams be... Then the expansion formula for the three frequencies is: (4) Where the subscript i represents the i-th set of raster images projected, and , This indicates the rounding operation.

4. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 1, characterized in that: Step 2.1 specifically includes the following steps: (5) in Let S be noise, and let S be a subset of L.

5. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 1, characterized in that: Step 2 also includes the following steps: Step 2.3: According to formula (6), the function and The integral becomes the sum of the entire region, that is: (7) Where S is a subset of the sensing data set L.

6. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 5, characterized in that: Step 2 also includes the following steps: Step 2.4: A simple regular operator The discrete representation of the image is as follows: (8)。 7. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 6, characterized in that: Step 2 also includes the following steps: Step 2.5: Based on formula (8), the discrete mathematical expression of the second-order regularization operator of the image is derived as follows: (9)。 8. The method for acquiring three-dimensional data based on galvanometer structured light according to claim 7, characterized in that: Step 2 also includes the following steps: Step 2.6: According to formula (9), in order to obtain Iterative optimization is performed using gradient descent: (10) Where k is the number of iterations.

Images