Resonance driving type scanner track detection method based on reconstruction image optimization

Abstract

The invention belongs to the technical field of scanning imaging, and discloses a resonance driving type scanner track detection method based on reconstruction image optimization, which comprises the following steps: initializing scanning track parameters of a resonance driving type scanner, and setting boundaries of the scanning track parameters; carrying out boundary correction on the current scanning track parameter, and substituting the current scanning track parameter into the track model to obtain a simulation track under the current iteration frequency; reconstructing an image by using the current simulation track and the acquired real-time imaging data, and calculating an image artifact evaluation index; judging whether a preset iteration termination condition is met or not, if not, taking the current image artifact evaluation index minimum as an optimization target, updating the current scanning track parameter and carrying out next iteration; and if yes, outputting the simulation track under the current iteration number and the corresponding reconstructed image, and performing affine deformation repair on the simulation track and the reconstructed image. According to the invention, the detection precision of the scanning track and the quality of the reconstructed image can be improved, and the implementation complexity and cost are reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of scanning imaging, and more specifically, relates to a method for detecting the trajectory of a resonance-driven scanner based on the optimization of a reconstructed image. Background Art

[0002] Resonance-driven scanners such as fiber optic scanners or MEMS galvanometers have the advantages of simple drive control, continuous scanning trajectories, and high scanning speeds, and are important components of microscopical imaging devices such as confocal endoscopes. However, resonance-driven scanners are sensitive to environmental and structural characteristics. When the environment changes or the scanner mechanism ages, the scanning trajectory will change. If the scanner trajectory is not re-corrected, the image quality will degrade.

[0003] In the prior art, the scanning imaging signal can be physically encoded by an aperture diaphragm mirror, and the frequency, phase, and amplitude of the scanner can be obtained by decoding the high level in the imaging signal. Then, the scanning trajectory can be corrected through a negative feedback circuit to make it consistent with the target trajectory, and a distortion-free image can be reconstructed. This method requires relying on hardware devices, and the additional feedback circuit increases the complexity and cost of system implementation. There is also a solution that, based on the fact that the pixel interleaving artifacts generated by the trajectory phase drift will increase the high-frequency components in the image, proposes to use the sum of the image frequency domain power spectra as an evaluation index, and uses a traversal algorithm to find the optimal phase for reconstructing the trajectory, thereby completing the image artifact repair. This method using a single-variable traversal optimization algorithm can only optimize the phase in one direction and is not applicable to the occasion where the phases in two orthogonal directions need to be optimized simultaneously. Moreover, the influence of the amplitude factor is not considered, resulting in a limited applicable range and low detection accuracy of the scanning trajectory, and thus the accuracy of the image reconstructed based on the detected trajectory is not high.

[0004] In addition, mechanical cross-coupling refers to the non-planar resonance phenomenon in which the vibration system generates coupled vibrations in the orthogonal direction of the excitation direction. For a biaxial resonance-driven scanner, the influence of its mechanical coupling phenomenon on the scanning trajectory can be eliminated through direct correction. However, for a uniaxial resonance-driven scanner, especially under its Lissajous scanning trajectory, the mechanical cross-coupling problem will have a serious impact on the scanning trajectory and cannot be eliminated through direct correction. Although the Lissajous scanning trajectory of a uniaxial resonance-driven scanner with mechanical cross-coupling is also periodically repeated, it does not conform to the mathematical description of the standard Lissajous trajectory. In practical applications, due to the limitations of manufacturing processes, assembly accuracy, and volume, etc., the mechanical cross-coupling phenomenon of a Lissajous micro scanner is difficult to avoid. For this kind of Lissajous scanning trajectory with mechanical cross-coupling, the direct detection method is generally used, such as directly measuring the timing trajectory by using sensors such as a position-sensitive detector (PSD). This method not only increases the system cost and volume, but also is restricted by factors such as system errors, detection accuracy, and noise, and the image reconstruction effect is not ideal. Summary of the Invention

[0005] In view of the above defects or improvement requirements of the prior art, the present invention provides a method for detecting the trajectory of a resonance-driven scanner based on the optimization of reconstructed images, aiming to obtain the high-precision scanning trajectory of the resonance-driven scanner in real time and reduce the implementation complexity and cost.

[0006] To achieve the above object, the present invention provides a method for detecting the trajectory of a resonance-driven scanner based on the optimization of reconstructed images, including:

[0007] S1. Initialize the scanning trajectory parameters of the resonance-driven scanner and set the boundaries of the scanning trajectory parameters; wherein, the scanning trajectory parameters include the phase and amplitude components of the scanning trajectory in the orthogonal direction of the scanner.

[0008] S2. After correcting the boundaries of the current scanning trajectory parameters, substitute them into the trajectory model to obtain the simulated trajectory at the current iteration number i; reconstruct the image based on the current simulated trajectory and the acquired real-time imaging data.

[0009] S3. Calculate the artifact evaluation index s of the current reconstructed image i , where the artifact evaluation index s i is used to measure the significance of the image interleaving artifacts caused by inaccurate simulated trajectories.

[0010] S4. Determine whether the preset iteration termination condition is reached. If not, optimize the current scanning trajectory parameters with the minimum of the artifact evaluation index s i as the optimization goal, let i = i + 1, and jump to S2; if so, output the simulated trajectory and the corresponding reconstructed image at the current iteration number.

[0011] Further, under the Lissajous scan of the resonance-driven scanner, the trajectory model is:

[0012] x(t) = a x sin(2πf 1 t + θ 1x ) + b x sin(2πf 2 t + θ 2x )

[0013] y(t) = b y sin(2πf 1 t + θ 1y ) + a y sin(2πf 2 t + θ 2y )

[0014] Among them, x(t) and y(t) represent the trajectory coordinates at time t, and t is the sampling time of the imaging device where the scanner is located; f 1 , f 2 are the resonance frequencies of the scanner in the x-direction and y-direction under Lissajous scanning, and the x-direction and y-direction form the orthogonal directions of the scanner; a x and a y are the amplitudes of the scanner in the x-direction and y-direction respectively, and b y and b x are the amplitudes of the coupled vibration generated by the scanner in the x-direction and the amplitude of the coupled vibration generated in the y-direction respectively; θ 1x and θ 2y are the phases of the scanner in the x-direction and y-direction respectively, and θ 1y and θ 2x are the phases of the coupled vibration generated by the scanner in the x-direction and the phase of the coupled vibration generated in the y-direction respectively.

[0015] Further, the artifact evaluation index s i is:

[0016]

[0017] Among them, M and N are the number of rows and columns of the reconstructed image respectively, and u and v are the coordinates in the frequency domain; |F i (u, v)| represents the frequency domain amplitude spectrum obtained by taking the modulus after the discrete Fourier transform of the reconstructed image f(x, y) under the i-th iteration, and D u,v represents the weight matrix;

[0018] Or, the artifact evaluation index s i is:

[0019]

[0020] Among them, p(r) represents the distribution ratio of the r-th gray level in the gray level set of the frequency domain amplitude spectrum |F i (u, v)|.

[0021] Further, after S4, it also includes affine deformation repair of the output simulated trajectory and the corresponding reconstructed image, specifically including:

[0022] Preset a reference object with a known shape in the imaging system where the scanner is located, calculate the linear transformation matrix corresponding to the affine deformation according to the field of view before imaging and the field of view after imaging of the reference object; linearly transform the simulated trajectory and the reconstructed image respectively using the linear transformation matrix to obtain the simulated trajectory and reconstructed image after affine deformation repair.

[0023] Further, the reference object with a known shape is a target or a diaphragm that limits the imaging field of view, and the target or the diaphragm is disposed in front of the imaging objective or behind the imaging objective.

[0024] Further, when reconstructing an image using the current simulated trajectory and the acquired real-time imaging data, the image reconstruction algorithm adopts a grid interpolation algorithm; the interpolation strategy of the grid interpolation algorithm is nearest neighbor interpolation, linear interpolation, quadratic spline interpolation, or natural interpolation.

[0025] Further, a multi-objective optimization algorithm is used to optimize the current scanning trajectory parameters, and the multi-objective optimization algorithm is a swarm optimization algorithm or a stochastic gradient descent algorithm with global optimization ability;

[0026] The iteration termination condition is determined by the optimization strategy of the multi-objective optimization algorithm, including that the artifact evaluation index s i has no change within a set threshold range, reaches a preset maximum number of iterations, or an outlier appears in the current trajectory.

[0027] The present invention also provides a resonance-driven scanner trajectory detection system based on reconstructed image optimization, including a computer-readable storage medium and a processor;

[0028] The computer-readable storage medium is used to store executable instructions;

[0029] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the resonance-driven scanner trajectory detection method based on reconstructed image optimization described in any one of the above.

[0030] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the resonance-driven scanner trajectory detection method based on reconstructed image optimization described in any one of the above.

[0031] The present invention also provides a computer program product, including a computer program, and when the computer program runs on a computer, it causes the computer to execute the resonance-driven scanner trajectory detection method based on reconstructed image optimization described in any one of the above.

[0032] Generally speaking, through the above technical solutions conceived by the present invention, the following beneficial effects can be achieved:

[0033] (1) The resonance-driven scanner trajectory detection method based on reconstructed image optimization of the present invention realizes imaging repair (image reconstruction) by modifying the trajectory, and the imaging repair and trajectory detection are completed synchronously. During the trajectory detection and image reconstruction processes, a multi-iteration optimization method is adopted. With the minimum of the artifact evaluation index of the current image as the optimization goal, the phase and amplitude components of the scanning trajectory in the orthogonal directions of the scanner are optimized. By means of real-time imaging data, a high-precision scanning trajectory of the resonance-driven scanner is obtained. Compared with other trajectory detection methods, the trajectory detection technology of the present invention does not rely on hardware detection equipment, is simple to implement, has high detection accuracy, good real-time performance, and low cost.

[0034] (2) Preferably, in the present invention, considering that for a single-axis resonance-driven scanner under a Lissajous scanning trajectory, the mechanical coupling phenomenon cannot be corrected due to system limitations, a Lissajous scanning trajectory model considering mechanical coupling is provided. This trajectory model is obtained based on the vibration equation of the vibration system under excitation and boundary conditions when the scanner vibrates. The present invention analyzes the vibration equation of the vibration system under excitation and boundary conditions when the scanner vibrates. Two resonance frequencies of the scanner in the orthogonal directions will respectively generate corresponding coupling terms b y sin(2πf 1 t + θ 1y ) and b x sin(2πf 2 t + θ 2x ). Thus, a Lissajous scanning trajectory model considering mechanical coupling is constructed. This trajectory model simultaneously considers eight optimization parameters including the amplitudes and phases of the scanner in the orthogonal directions under Lissajous scanning and the amplitudes and phases of the corresponding coupling terms. Furthermore, based on the iterative optimization method, the trajectory detection is indirectly completed by means of the imaging function without the need for additional hardware such as a position-sensitive detector (PSD), reducing the system cost, shrinking the volume, and having a wider application range, especially being applicable to a narrow body cavity environment; moreover, the detected trajectory is closer to the real trajectory, greatly reducing the requirements for the design and machining and assembly accuracy of the scanner. In addition, the correction of the synchronous clock error is also achieved while optimizing the four phase terms.

[0035] (3) Preferably, the artifact evaluation index constructed based on the average value of the dot product of the frequency-domain amplitude spectrum of the current reconstructed image and the weight matrix considers the frequency distribution ratio information of the image frequency-domain amplitude spectrum. Compared with directly obtaining the amplitude average value, when optimizing with the minimum of this artifact evaluation index as the optimization goal, the convergence region near the optimal value is larger, which is beneficial to improving the optimization speed. The artifact evaluation index constructed based on information entropy considers the intensity ratio information of the image frequency-domain amplitude spectrum and is less affected by the image structure, being more suitable for the case where the imaging target structure is complex.

[0036] (4) Further, considering that the artifact evaluation index constructed in the present invention cannot evaluate the field-of-view deformation of the image, and during the optimization process, it is found that the constructed evaluation index will also introduce a certain degree of field-of-view deformation. Therefore, after completing the image artifact repair, it further includes solving the linear transformation matrix of the field-of-view deformation based on the field-of-view prior information, and completing the affine deformation repair of the simulated trajectory and the reconstructed image, so as to further improve the detection accuracy of the scanning trajectory and the quality of the reconstructed image.

[0037] (5) Preferably, in S4, considering that the present invention is for multi-parameter optimization and the image information interferes with the image degradation index, local optimality may occur in the optimization process. Therefore, the multi-objective optimization algorithm selects the swarm optimization algorithm or the stochastic gradient descent algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the resonance-driven scanner trajectory detection method based on reconstructed image optimization in the embodiment of the present invention.

[0039] Figure 2 It is a flowchart of the resonance-driven scanner trajectory detection method based on reconstructed image optimization in the embodiment of the present invention.

[0040] Figure 3 It is the ideal trajectory of the Lissajous micro scanner and the non-ideal trajectory generating mechanical cross-coupling.

[0041] Figure 4 It is the imaging result obtained during the process of applying the method in the embodiment of the present invention to a fiber optic scanning confocal endoscope to realize image degradation repair. The numbers 1-8 in the figure represent Figure 4 the serial numbers of the sub-images. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0043] Embodiment 1

[0044] As Figures 1 - 2 shown, the embodiment of the present invention provides a resonance-driven scanner trajectory detection method based on reconstructed image optimization, which mainly includes:

[0045] S1. Initialization of scanning trajectory model parameters: Randomly or based on prior values, initialize the scanning trajectory parameters of the resonance-driven scanner, set the boundaries of the scanning trajectory parameters, and preset the maximum number of iterations MaxIteration; among them, the scanning trajectory parameters include the phase and amplitude components of the scanning trajectory in the orthogonal direction of the scanner.

[0046] S2. Reconstruct the image based on the simulated trajectory: After correcting the boundaries of the current scanning trajectory parameters, substitute them into the trajectory model, and after normalization, obtain the simulated trajectory at the current iteration number i; reconstruct the image based on the current simulated trajectory and the acquired real-time imaging data.

[0047] S3. Calculate the evaluation index of image artifacts: Calculate the artifact evaluation index s of the current reconstructed image i ; where the artifact evaluation index is a statistical index of the image frequency domain information, used to measure the significance of the image aliasing artifacts caused by inaccurate simulated trajectories.

[0048] S4. Iteratively optimize the simulated trajectory parameters: Determine whether the preset iteration termination condition is reached. If not, take the minimum of the current artifact evaluation index as the optimization goal, use a multi-objective optimization algorithm to update the current scanning trajectory parameters, let i = i + 1, and jump to S2; if so, output the simulated trajectory and the corresponding reconstructed image at the current iteration number, and complete the pixel aliasing artifact repair.

[0049] As a further design of the present invention, after completing the image artifact repair, it further includes:

[0050] S5. Perform affine deformation repair on the output simulated trajectory and the corresponding reconstructed image to obtain the simulated trajectory and the corresponding reconstructed image after affine deformation repair.

[0051] Preferably, in S1, the boundaries of the trajectory parameters refer to the value range of the trajectory parameters, which are set according to the actual empirical values of the trajectory model, used to exclude unreasonable situations and reduce the calculation cost.

[0052] Those skilled in the art know that in S2, boundary correction is required after each iteration of the trajectory parameter update. When the amplitude in the updated trajectory parameters exceeds the boundary value, make it equal to the boundary value. When the phase exceeds the boundary value, make it ±2kπ (k is an integer), and convert it to the equivalent phase value within the phase value range.

[0053] The trajectory model is constructed based on the vibration characteristics of the scanner. Optionally, the trajectory model can be a Lissajous scanning trajectory model, a spiral scanning trajectory model, etc.

[0054] Preferably, in the embodiment of the present invention, a Lissajous scanning trajectory model considering mechanical coupling is provided, and the formula is as follows:

[0055] x(t) = ax sin(2πf 1 t + θ 1x ) + b x sin(2πf 2 t + θ 2x )

[0056] y(t) = b y sin(2πf 1 t + θ 1y ) + a y sin(2πf 2 t + θ 2y )

[0057] where x(t) and y(t) represent the trajectory coordinates at time t, and t is the sampling time of the imaging device where the scanner is located; f 1 , f 2 are two resonance frequencies of the scanner in the orthogonal directions under Lissajous scanning, that is, f 1 is the resonance frequency of the scanner in the x - direction and f 2 is the resonance frequency of the scanner in the y - direction. The x - direction and the y - direction form two mutually orthogonal directions of the scanner; a x , b x , a y , b y , θ 1x , θ 1y , θ 2x , θ 2y are 8 parameters to be optimized related to the driving load, mechanical characteristics of the scanner, boundary conditions, etc. a x and a y are the amplitudes of the scanner in the x - direction and the y - direction respectively, and b y and b x are the amplitudes of the coupled vibration generated in the x - direction and the amplitude of the coupled vibration generated in the y - direction respectively; θ 1x and θ 2y are the phases of the scanner in the x - direction and the y - direction respectively, and θ 1y and θ 2x are the phases of the coupled vibration generated in the x - direction and the phase of the coupled vibration generated in the y - direction respectively; that is, b y sin(2πf 1 t + θ 1y ) is the coupled vibration generated by the scanner in the x - direction, and b x sin(2πf 2 t + θ 2x ) is the coupled vibration generated by the scanner in the y - direction; Each parameter update requires domain - constraint correction. The ideal trajectory of the Lissajous micro - scanner and the non - ideal trajectory that generates mechanical cross - coupling are asFigure 3 as shown

[0058] In an embodiment of the present invention, the resonance-driven scanner is regarded as a stable and continuous vibration system. Based on the vibration equation of the scanner, the scanning trajectory model is solved. The position-sensitive detector is used to collect the actual scanning trajectories of the scanner under different working conditions (changing the excitation, structural parameters, etc.) of the scanner. The trajectory model is used to fit each set of trajectory data, and the correctness of the trajectory model is verified through correlation analysis.

[0059] In S2, normalizing the trajectory is for facilitating image reconstruction. After translating the trajectory and scaling each coordinate (x, y) on the trajectory proportionally, the operation is such that the overall maximum value in the generated simulated trajectory equals 1 and the minimum value equals 0.

[0060] In S2, the real-time imaging data is a one-dimensional grayscale signal obtained by isochronous sampling of the imaging device where the scanner is located. In the embodiment of the present invention, the imaging target itself has structural or texture information, and in order to ensure the accuracy of the trajectory detection effect, the imaging result needs to have a certain signal-to-noise ratio.

[0061] Preferably, in S2, it further includes preprocessing the real-time imaging data, including: frame interception (intercepting the real-time imaging data into each frame of data), noise reduction, normalization and other operations.

[0062] Specifically, in S2, using the current simulated trajectory and the obtained real-time imaging data, an image reconstruction algorithm is used to reconstruct the image. Preferably, the image reconstruction algorithm adopts a grid interpolation algorithm, and the interpolation strategy can be selected from nearest neighbor interpolation, linear interpolation, quadratic spline interpolation, natural interpolation, etc.

[0063] Preferably, in S3, the frequency domain amplitude spectrum obtained by performing discrete Fourier transform and taking the modulus of the current reconstructed image is calculated, and the average value of the dot product of the frequency domain amplitude spectrum and the weight matrix is used as the current image artifact evaluation index s i ; wherein, the weight matrix is composed of the sum of the p-th powers of the distances from each point in the frequency domain amplitude spectrum to the horizontal and vertical axes of the spectrum center, and after taking the negative value; the corresponding calculation formula is:[[]]

[0064]

[0065] wherein, M and N are respectively the number of rows and columns of the reconstructed image, and u and v are the coordinates in the frequency domain; |F i (u, v)| represents the frequency domain amplitude spectrum obtained by performing discrete Fourier transform and taking the modulus of the reconstructed image f(x, y) in the i-th iteration, and the calculation formula is as follows:[[]]

[0066]

[0067] wherein, Re(F i(u, v)) and Im(F i (u, v)) are the real part and the imaginary part of F i (u, v) respectively; F i (u, v) is the discrete Fourier transform of the reconstructed image f(x, y).

[0068] D u,v represents the weight matrix, which is formed by taking the negative value of the sum of the p-th powers of the distances from each point (u, v) in the frequency-domain amplitude spectrum to the horizontal and vertical axes of the spectrum center. The smaller the value, the more low-frequency components in the frequency domain.

[0069]

[0070] Preferably, p can be taken as 1.

[0071] As a preference, the artifact evaluation index s in the embodiments of the present invention i can also be expressed as:

[0072]

[0073] where p(r) represents the distribution ratio of the r-th gray level in the gray level set of the frequency-domain amplitude spectrum |F i (u, v)|.

[0074] As a preference, in S4, considering that there are multiple parameters to be optimized in the present invention and the image information interferes with the image degradation index, local optimality may occur in the optimization process. Therefore, for the multi-objective optimization algorithm, a swarm optimization algorithm or a stochastic gradient descent algorithm with global optimization ability is selected.

[0075] The iteration termination condition is determined according to the selected global optimization algorithm, including that the image artifact evaluation index s i has no change within the set threshold range, reaches the preset maximum number of iterations, or an outlier appears in the current trajectory, etc.

[0076] As a further design of the present invention, the image artifact evaluation index constructed in the present invention cannot evaluate the field-of-view deformation of the image, and during the optimization process, it is found that the constructed evaluation index will also introduce a certain degree of field-of-view deformation. Therefore, after completing the image artifact repair, an affine deformation repair process is also included. Affine deformation repair refers to performing an affine transformation (rotation, shear, and scaling transformation) on the optimized trajectory and image to repair the overall geometric deformation of the field of view, and this kind of deformation will not generate interleaved artifacts.

[0077] In S5, affine deformation repair is performed on the output simulated trajectory and the corresponding reconstructed image, including:

[0078] A reference object with a known shape is preset in the imaging system, and the linear transformation matrix corresponding to the affine deformation is calculated according to the field of view before imaging and the field of view after imaging of the reference object; the output simulated trajectory and the corresponding reconstructed image are respectively linearly transformed by using the linear transformation matrix to obtain the simulated trajectory and the reconstructed image after affine deformation repair.

[0079] Preferably, the reference object with a known shape is a target or a diaphragm that limits the imaging field of view, and the target or the diaphragm is arranged in front of the imaging objective lens or behind the imaging objective lens.

[0080] Considering that the image field of view after artifact repair in the present invention is generally an oblique ellipse, the reference object with a known shape in the present invention is preferably a circular diaphragm, which is arranged in front of the imaging objective lens. The linear transformation matrix corresponding to the affine deformation is calculated based on the elliptical field of view after imaging of the circular diaphragm and the circular field of view before imaging, and the output simulated trajectory and the corresponding reconstructed image are respectively linearly transformed by using the linear transformation matrix to obtain the simulated trajectory and the reconstructed image after affine deformation repair.

[0081] Next, the method of the present invention is applied to a fiber optic scanning confocal endoscope system, and the imaging sample is a resolution test chart, and the resolution test chart also serves as a reference object with a known shape.

[0082] (1) Imaging data preprocessing: The excitation signals of the scanner are known information, with frequencies of f 1 = 2408Hz, f 2 = 2304Hz, and the imaging parameters are set as a frame rate of 8fps, a sampling rate of 10MHz, and a single-frame data length of 1.25M samples. After obtaining a single-frame data, sliding window Gaussian filtering and normalization are sequentially completed to obtain normalized imaging data.

[0083] (2) Parameter initialization: When the reconstruction result of the pre-stored trajectory degrades severely or there is no pre-stored trajectory, 8 initial parameter values of the trajectory model can be randomly generated. According to experience, the two frequencies respectively account for the main components in the orthogonal directions. It is specified that a x , a y are the main components, and the value range is [0.9, 1]; b x , b y are the secondary components (coupling components), and the value range is [0, 0.2], and the value ranges of θ 1x , θ 1y , θ 2x , θ 2y are [0, 2π]. There is no pre-stored trajectory in this correction, so a set of trajectory parameters is randomly generated, and the reconstruction result is Figure 4As shown in the sub - figure numbered 1 in [reference], it can be seen that when there is no iteration, the degradation phenomenon of the reconstructed graph is serious. The trajectory parameters randomly generated or updated iteratively later need boundary correction. In contrast, when the degradation of the reconstruction result of the pre - stored trajectory is not serious, the pre - stored scanning trajectory can be used as the initial value, which can reduce the calculation time. And the maximum number of iterations MaxIteration = 1000 is set.

[0084] (3) Substitute into the mechanical coupling trajectory formula of the Lissajous micro - scanner to generate a simulated trajectory, and then perform normalization processing.

[0085] (4) Adopt a general grid interpolation algorithm, use the nearest - neighbor interpolation strategy for the interpolation strategy, and reconstruct the degraded image; and calculate the artifact evaluation index of the current image.

[0086] (5) Use the update strategy of the Adam optimization algorithm to iteratively update the trajectory parameters, and judge the loop termination conditions: the artifact evaluation index of the image after 50 iterations does not change within the set threshold range and reaches the maximum number of iterations 1000. If either condition is met, the loop can be exited, and the simulated trajectory and the corresponding reconstructed image at the current iteration are output to complete the image artifact repair. During the algorithm correction process, several intermediate optimization processes are selected and shown as Figure 4 in the sub - figures numbered 2 - 6 in [reference]. It can be seen that compared with the sub - figure numbered 1, the imaging is clearer. Calculate the inverse matrix of the linear transformation through the prior information of the resolution test chart, and then apply it to Figure 4 the sub - figure numbered 7 in [reference], and obtain the imaging result of the resolution test chart as shown in Figure 4 the sub - figure numbered 8 in [reference]. The results show that the method of the present invention can realize the function of detecting the real trajectory of the scanner, and the reconstructed image has a good effect.

[0087] Embodiment 2

[0088] The embodiment of the present invention provides a resonance - driven scanner trajectory detection system based on reconstructed image optimization, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it realizes the steps of the resonance - driven scanner trajectory detection method based on reconstructed image optimization in the above - mentioned Embodiment 1.

[0089] The related technical solutions are the same as above and will not be elaborated here.

[0090] Embodiment 3

[0091] The embodiment of the present invention provides a computer - readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the steps of the resonance - driven scanner trajectory detection method based on reconstructed image optimization in the above - mentioned Embodiment 1.

[0092] Specifically, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0093] The related technical solutions are the same as above and will not be elaborated here.

[0094] Embodiment 4

[0095] The embodiment of the present application provides a computer program product, including a computer program, which, when running on a computer, causes the computer to execute the steps of the resonance-driven scanner trajectory detection method based on reconstructed image optimization in Embodiment 1 above.

[0096] The related technical solutions are the same as above and will not be elaborated here.

[0097] Those skilled in the art can easily understand that the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A resonance driven scanner trajectory detection method based on reconstructed image optimization, characterized in that: include: S1, initializing scanning trajectory parameters of the resonance driven scanner and setting the boundaries of the scanning trajectory parameters; wherein the scanning trajectory parameters include phase and amplitude components of the scanning trajectory in the orthogonal direction of the scanner; S2, after boundary correction of the current scanning trajectory parameters, the parameters are brought into the trajectory model to obtain the simulated trajectory under the current iteration number i; and the image is reconstructed based on the current simulated trajectory and the acquired real-time imaging data; S3. Calculate the artifact evaluation index s of the current reconstructed image i , wherein the artifact evaluation index s i Used to measure the significance of image interlacing artifacts caused by inaccurate simulation trajectories; S4, determine whether the preset iteration termination condition is reached, if not, use the artifact evaluation index s i Minimum is the optimization target, optimize the current scanning trajectory parameters, set i=i+1, and jump to S2; if so, output the simulation trajectory and the corresponding reconstructed image under the current number of iterations.

2. The resonance driven scanner trajectory detection method based on reconstructed image optimization according to claim 1, characterized in that: The trajectory model of the resonant driven scanner under Lissajous scanning is: x(t)=a x sin(2πf1t+θ 1x )+b x sin(2πf2t+θ 2x ) y(t)=b y sin(2πf1t+θ 1y )+a y sin(2πf2t+θ 2y ) Wherein, x(t) and y(t) represent the trajectory coordinates at time t, and t is the sampling time of the imaging device where the scanner is located; f1 and f2 are the resonant frequencies of the scanner in the x direction and y direction under Lissajous scanning, and the x direction and y direction constitute the orthogonal direction of the scanner; a x and a y are the amplitudes of the scanner in the x and y directions, respectively, and b y and b x are the amplitude of the coupled vibration generated by the scanner in the x direction and the amplitude of the coupled vibration generated in the y direction respectively; θ 1x and θ 2y are the phases of the scanner in the x and y directions, θ 1y and θ 2x They are respectively the phase of the coupled vibration generated by the scanner in the x-direction and the phase of the coupled vibration generated in the y-direction.

3. The resonance driven scanner trajectory detection method based on reconstructed image optimization according to claim 1 or 2, characterized in that: The artifact evaluation index s i for: Where M and N are the number of rows and columns of the reconstructed image, u and v are the coordinates in the frequency domain; |F i (u,v)| represents the frequency domain amplitude spectrum obtained by discrete Fourier transform and modulus of the reconstructed image f(x,y) at the i-th iteration, D u,v represents the weight matrix; Alternatively, the artifact evaluation index s i for: Wherein, p(r) represents the frequency domain amplitude spectrum |F i The distribution ratio of the rth gray level in the gray level set of (u,v)|.

4. The resonance driven scanner trajectory detection method based on reconstructed image optimization according to claim 3, characterized in that: After S4, the output simulation trajectory and the corresponding reconstructed image are also subjected to affine deformation repair, which specifically includes: A reference object of known shape is preset in the imaging system where the scanner is located, and a linear transformation matrix corresponding to the affine deformation is calculated according to the field of view of the reference object before imaging and the field of view after imaging; the simulated trajectory and the reconstructed image are linearly transformed using the linear transformation matrix respectively to obtain the simulated trajectory and the reconstructed image after the affine deformation is repaired.

5. The resonance driven scanner trajectory detection method based on reconstructed image optimization according to claim 4, characterized in that: The reference object with a known shape is a target or an aperture that limits the imaging field of view, and the target or the aperture is arranged in front of the imaging objective lens or behind the imaging objective lens.

6. The resonance driven scanner trajectory detection method based on reconstructed image optimization according to claim 1 or 5, characterized in that: When reconstructing an image using the current simulation trajectory and the acquired real-time imaging data, the image reconstruction algorithm adopts a grid interpolation algorithm; the interpolation strategy of the grid interpolation algorithm is nearest neighbor interpolation, linear interpolation, quadratic spline interpolation or natural interpolation.

7. The resonance driven scanner trajectory detection method based on reconstructed image optimization according to claim 1 or 5, characterized in that: A multi-objective optimization algorithm is used to optimize the current scanning trajectory parameters, wherein the multi-objective optimization algorithm is a group optimization algorithm or a stochastic gradient descent algorithm with global optimization capability; The iteration termination condition is determined by the optimization strategy of the multi-objective optimization algorithm, including the artifact evaluation index s after multiple iterations. i There is no change within the set threshold range, the preset maximum number of iterations is reached, or an outlier appears in the current trajectory.

8. A resonance driven scanner trajectory detection system based on reconstructed image optimization, characterized in that: comprising a computer readable storage medium and a processor; The computer-readable storage medium is used to store executable instructions; The processor is used to read the executable instructions stored in the computer-readable storage medium to execute the resonance-driven scanner trajectory detection method based on reconstructed image optimization according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the resonance-driven scanner trajectory detection method based on reconstructed image optimization as described in any one of claims 1 to 7 is implemented.

10. A computer program product, characterized in that The invention comprises a computer program, which, when running on a computer, enables the computer to execute the resonance driven scanner trajectory detection method based on reconstructed image optimization as claimed in any one of claims 1 to 7.

Images