Shack-hartmann wavefront sensor wavefront recovery method based on image enhancement

Abstract

The application discloses a Shack-Hartmann wavefront sensor wavefront recovery method based on image enhancement, and is influenced by factors such as atmospheric turbulence intensity, transmission distance, beacon light return light characteristics, and limited dynamic range of a detector, uneven near-field light intensity distribution and dynamic change of a to-be-measured light beam, signal and noise mixing, and unclear conditions of a wavefront sensor image, so that a sub-spot centroid, that is, a wavefront slope calculation error is increased, thereby reducing the recovery accuracy of the wavefront sensor. In view of the above problems, the application firstly uses an image enhancement technology to perform image enhancement on the wavefront sensor image, highlights a low-frequency sub-band of a central signal, suppresses high-frequency noise in the four surrounding areas, improves image quality, reduces the influence of slope calculation errors of a non-uniform light intensity sub-aperture on global wavefront recovery, and then can improve the sub-aperture slope extraction and wavefront detection accuracy of the Shack-Hartmann wavefront sensor.

Description

Technical Field

[0001] This invention belongs to the field of wavefront detection technology, and in particular relates to a wavefront restoration method for Shaker-Hartmann wavefront sensors based on image enhancement, which can be used for wavefront detection in multiple fields such as astronomical observation, laser atmospheric transmission compensation, and free-space laser communication. Background Technology

[0002] The Shack-Hartmann wavefront sensor (SHWFS) is widely used in adaptive optics systems due to its simple principle, high light energy utilization, and high speed. The SHWFS mainly consists of a microlens array and a photodetector. When the incident beam has wavefront distortion, the wavefront tilt within the sub-aperture range will cause spot drift. By measuring the offset of the centroid of the focused sub-spot relative to the calibration position, the wavefront slope corresponding to that sub-aperture can be obtained. After obtaining the incident wavefront slope data, the phase distribution of the incident wavefront can be obtained through a wavefront reconstruction algorithm.

[0003] In practical applications, affected by factors such as atmospheric turbulence intensity, transmission distance, and beacon light backlight characteristics, the near-field light intensity distribution of the incident beam of the Shack-Hartmann wavefront sensor is uneven and dynamically changes. The wavefront sensor has a low signal-to-noise ratio region, and the signal in some regions is buried in noise. At this time, the sub-aperture slope calculation error increases, which reduces the restoration accuracy of the wavefront sensor. Ultimately, this leads to a decrease in the correction effect of the adaptive optics system, and even a closed-loop instability. At present, the main methods used are windowed threshold weighted centroid method ([1] Wei Ping. Research on image signal processing method of Hartmann wavefront sensor under low signal-to-noise ratio conditions [D]. University of Electronic Science and Technology of China, 2021.) and cross-correlation algorithm (E. Sidick, JJ Green, RMMorgan, et al. Adaptive cross-correlation algorithm for extended scene Shack-Hartmann wavefront sensing [J]. Optics Methods such as Letters, 2008, 33(3):213-215, Local Adaptive Thresholding (Li Xuxu, Li Xinyang, Wang Caixia. Local Adaptive Thresholding Method for Sub-Aperture Spots of Hartmann Sensor [J]. Opto-Electronic Engineering, 2018, 45(10):170699), and Centroid Localization Based on Deep Learning (Li Ziqiang. Adaptive Optical Wavefront Sensing Technology Based on Deep Learning [D]. University of Chinese Academy of Sciences (Institute of Opto-Electronics, Chinese Academy of Sciences), 2021) attempt to improve the centroid calculation error of low signal-to-noise ratio sub-spots under uneven light intensity distribution, but cannot completely avoid the impact of these weak spot data errors on the full-aperture wavefront reconstruction accuracy.

[0004] Therefore, when the near-field light intensity distribution of the incident beam is uneven and dynamically changing, it is necessary to find a method that can reduce the impact of the calculation error of the centroid of the dim and weak sub-spot with low signal-to-noise ratio on the wavefront reconstruction accuracy, improve the wavefront detection accuracy under the dynamic fluctuations of the near-field of the incident beam, and further expand the adaptability of the adaptive optics system under conditions such as strong turbulence and long transmission distance. Summary of the Invention

[0005] The technical problem to be solved by this invention is to provide a wavefront restoration method for Shak-Hartmann wavefront sensors based on image enhancement, thereby solving the problem of high-precision wavefront restoration of Shak-Hartmann wavefront sensors under non-uniform near-field light intensity distribution of incident beams.

[0006] The technical solution adopted by this invention to solve the above-mentioned technical problems is: a wavefront restoration method for a Shaker-Hartmann wavefront sensor based on image enhancement. This method enhances the image based on the initial light intensity of the sub-spot of the Shaker-Hartmann wavefront sensor, highlighting central information that is beneficial for centroid extraction, suppressing uninteresting parts, making the differences between different features more obvious, improving the signal-to-noise ratio of the sub-spot, and restoring the wavefront using the enhanced sub-aperture information. This method is specifically implemented through the following steps:

[0007] Step 1: Based on the sub-aperture arrangement and segmentation design, set the total number of sub-apertures to m, acquire images from a Shaker-Hartmann wavefront sensor, and decompose the sub-apertures into low-frequency and high-frequency information. Specifically, use an ideal low-pass filter h and an ideal high-pass filter g to perform directional decomposition on the image, and then apply fixed-coefficient attenuation and nonlinear transformation to the high-frequency and low-frequency components respectively to achieve enhancement and improve the signal-to-noise ratio, as shown in the equation: Where c and d are gray-level feature parameters. The low-frequency coefficients of the enhanced low-frequency image, To enhance the low-frequency coefficients before enhancement, For the high-frequency coefficients of the enhanced high-frequency image, The high-frequency coefficients of the high-frequency image before enhancement are shown. The image size is M×N, k1 represents a fixed attenuation coefficient, and k2 represents a nonlinear enhancement operator. w(x,y) is a weighted kernel function. After enhancing the low-frequency and high-frequency components respectively, an inverse transform is performed to reconstruct the image, thus obtaining the enhanced image.

[0008] Step 2: Set the number of aberration modes to be restored to n, and calculate the centroid offset matrix D of all sub-aperture spots with each aberration mode as input. The dimension is 2m×n. The inverse matrix of D is the mode restoration matrix R generated by the sub-aperture, with a dimension of n×2m.

[0009] Step 3: Calculate the centroid data x of sub-spots from the enhanced spot array image using spot localization technology.c y c The slope vector S of the current probe wavefront is calculated using sub-spot offset, with a dimension of 2m×1. The aberration mode coefficients are obtained from A=R·S, where R is the aberration mode coefficient restoration matrix. Based on the aberration modes and their corresponding coefficients, the wavefront is reconstructed by summing the results.

[0010] Furthermore, the image enhancement method described in step 1 includes wavelet transform, low-pass and high-pass filtering, homomorphic filtering, and other image enhancement methods, or any other enhancement method that can achieve high-frequency and low-frequency decomposition of the image.

[0011] Furthermore, the spot localization technology described in step 3 includes spot localization methods such as windowing, weighted centroid method, threshold centroid method, matched filtering, and registration-based algorithms, or any other method that can locate the spot position.

[0012] Furthermore, the aberration modes described in steps 2 and 3 can be Zernike aberration modes, Legendre aberration modes, or any other two-dimensional complete orthogonal basis functions.

[0013] Compared with the prior art, the present invention has the following advantages:

[0014] (1) The present invention uses an image enhancement method to improve the pixel value of the low-frequency signal of the sub-spot image of the Hartmann wavefront sensor and suppress the noise around the sub-spot. Unlike the conventional image enhancement technology that processes the entire image, this invention enhances each sub-aperture image separately by splitting the sub-aperture, so that the non-uniform sub-spot can be independently decomposed and reconstructed, thereby improving the centroid positioning accuracy of the low signal-to-noise ratio sub-spot of the wavefront sensor, and thus improving the wavefront detection accuracy of the non-uniform near-field beam and the low signal-to-noise ratio sub-spot image.

[0015] (2) The present invention has high compatibility when the incident beam of the wavefront sensor is dynamically non-uniformly distributed in the near field. It does not require real-time adjustment of the noise removal algorithm and the spot positioning algorithm parameters according to the state of the Shaker-wavefront sensor sub-spot and the noise characteristics of the detector. It can be widely applied to wavefront restoration under conditions such as strong skylight background and strong environmental noise.

[0016] (3) Sub-aperture segmentation enables multi-threaded parallel image enhancement, which shortens processing time and improves real-time performance. Attached Figure Description

[0017] Figure 1 The flowchart shows the wavefront restoration algorithm of the Shaker-Hartmann wavefront sensor based on wavelet transform image enhancement.

[0018] Figure 2 The images show the light spot patterns of the Shack-Hartmann wavefront sensor after noise reduction, noise enhancement, and image augmentation.

[0019] Figure 3 The peak signal-to-noise ratio increase after image enhancement for each sub-aperture;

[0020] Figure 4 For noise-free, noisy, and image enhancement-restored wavefronts and wavefront restoration residuals. Detailed Implementation

[0021] To make the objectives and technical solutions of this invention clearer, the invention will be further described in detail below with reference to specific implementation examples and the accompanying drawings.

[0022] In the example, the incident beam aperture is 30mm, the laser wavelength is 1064nm, the wavefront sensor has 14×14 sub-apertures with a sub-aperture size of 270μm, the microlens focal length is 11.5mm, the CCD detector pixel size is 15μm×15μm, and the bit depth is 10 bits.

[0023] like Figure 1 As shown, a method for wavefront restoration of a Shak-Hartmann wavefront sensor based on image enhancement includes the following steps:

[0024] Step 1: Based on the sub-aperture arrangement and segmentation design, the total number of sub-apertures is set to 156. Images are acquired using a Shaker-Hartmann wavefront sensor. For each sub-aperture, a wavelet transform image decomposition algorithm is applied to divide the original image into low-frequency and high-frequency information. Fixed-coefficient attenuation and nonlinear transformation are applied to the high-frequency and low-frequency components respectively to achieve enhancement and improve the signal-to-noise ratio, as shown in the equation: Where c and d are gray-level feature parameters. The low-frequency coefficients of the enhanced low-frequency image, To enhance the low-frequency coefficients before enhancement, For the high-frequency coefficients of the enhanced high-frequency image, For the high-frequency coefficients of the high-frequency image before enhancement, k1 is taken as a fixed attenuation coefficient of 0.75, and k2 represents the nonlinear enhancement operator. w(x,y) is the weighted kernel function, where the grayscale features are c=6 and d=11. After enhancing the low-frequency and high-frequency components separately, an inverse wavelet transform is performed for reconstruction to obtain the enhanced image. The low-frequency signal at the center of the sub-spot is enhanced, the surrounding high-frequency noise is suppressed, and the 156 sub-apertures are stitched together to obtain the final enhanced image.

[0025] Step 2: Extract the position coordinates (x, y) of the sub-spot in the x and y directions using sub-spot localization technology. c y c And calculate the offset Δx = x of the sub-spot relative to the calibration position. c -x0, Δy=y c-y0 (x0 and y0 are the centroid zero points of each sub-aperture of the Hartmann wavefront sensor, calibrated by using plane waves as input). A fixed threshold Thresh_fix is ​​subtracted from the entire Shaker-Hartmann wavefront sensor, and data with image values ​​less than 0 are set to zero. In this embodiment, the fixed threshold Thresh_fix is ​​50. The images corresponding to the sub-apertures of the Shaker-Hartmann wavefront sensor are traversed. For each individual sub-aperture, the image peak value is subtracted by Thresh_p times, and data with image values ​​less than 0 are set to zero. Thresh_p is a threshold coefficient, which is 0.2 in this embodiment. After removing image noise using the thresholding method, the sub-spot positions are calculated using the weighted centroid method.

[0026] Step 3: Generate the mode restoration matrix. Using the spot centroid offset data, calculate the aberration mode coefficients through the mode restoration matrix. Based on the aberration modes and their corresponding coefficients, sum them up to reconstruct the wavefront.

[0027] Step 3.1: Set the number of aberration patterns to be measured to N. z The value is 35, the aberration mode is Zernike aberration mode, and the sub-aperture spot centroid offset data matrix D is calculated with each Zernike aberration mode as input. Then the inverse matrix D' is the corresponding aberration mode coefficient restoration matrix R.

[0028] Step 3.2: Calculate the wavefront slope matrix S using sub-spot offset data. xy In the slope matrix, the slopes in the x and y directions of the sub-aperture are placed alternately, where the wavefront slope in the x direction is equal to 2πΔx / (λf), and the wavefront slope in the y direction is equal to 2πΔy / (λf), where λ is the wavelength of the incident beam and f is the focal length of the microlens.

[0029] Step 3.3: Based on the spot centroid offset data vector S, use the aberration mode coefficient restoration matrix R to calculate the 35th-order Zernike aberration mode coefficients in the wavefront distortion to be measured, i.e., the mode coefficient vector A = R·S. Based on the solved 35th-order Zernike aberration mode coefficients A and the corresponding aberration modes, sum them up to reconstruct the wavefront.

[0030] Figure 2 The image shows the light spot patterns of the Shak-Hartmann wavefront sensor with and without noise, as well as with image enhancement. Figure 3 The peak signal-to-noise ratio improvement after image enhancement for each sub-aperture is given. Figure 4 The wavefronts and wavefront restoration residuals are presented for noise-free, noisy, and image enhancement restorations. For example... Figure 2 As shown, wavelet transform coefficient enhancement can improve the pixel values ​​of a light spot array image and enhance the clarity of sub-spots in low-light conditions. Figure 3As shown, after image enhancement, the peak signal-to-noise ratio of 92% of the sub-apertures is improved, highlighting effective information and facilitating subsequent centroid extraction. Figure 4 As shown, the two-dimensional distribution of the wavefront restored using the noise-free sub-spot image and the image-enhanced sub-spot image is consistent. The wavefront restored using the noisy sub-spot image differs significantly from the above results. The RMS values ​​of the residuals of the wavefront restorations of the noise-free sub-spot image, the noisy sub-spot image, and the image-enhanced image are 0.0382λ, 0.1428λ, and 0.0758λ, respectively. Figure 4 The results show that, compared to directly calculating the spot position using the segmented threshold centroid method for wavefront reconstruction of noisy images, the method described in this invention improves wavefront reconstruction accuracy by approximately two times. In summary, this invention improves the wavefront detection accuracy of wavefront sensors by enhancing the sub-spot images of the Shaker-Hartmann wavefront sensor.

[0031] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any transformations or substitutions that can be conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of the present invention.

Claims

1. A wavefront restoration method for a Shaker-Hartmann wavefront sensor based on image enhancement, characterized in that: Image enhancement is performed based on the initial light intensity values ​​of the sub-spots of the Shaker-Hartmann wavefront sensor to highlight the center signal, suppress edge noise, improve the signal-to-noise ratio, and reconstruct the wavefront using the enhanced sub-aperture information. This method is specifically implemented through the following steps: Step 1: Based on the sub-aperture arrangement and segmentation design, set the total number of sub-apertures to m, acquire images from the Shaker-Hartmann wavefront sensor, segment the sub-apertures, and then perform image enhancement on each sub-aperture. The original image is divided into low-frequency signals and high-frequency noise. Then, fixed-coefficient attenuation and nonlinear transformation are applied to the high-frequency and low-frequency components respectively to achieve enhancement and improve the signal-to-noise ratio, as shown in the equation: , Where c and d are gray-level feature parameters, The low-frequency coefficients of the enhanced low-frequency image, To enhance the low-frequency coefficients before enhancement, For the high-frequency coefficients of the enhanced high-frequency image, The high-frequency coefficients of the high-frequency image before enhancement are shown. The image size is M×N, k1 represents a fixed attenuation coefficient, and k2 represents a nonlinear enhancement operator. w(x,y) is a weighted kernel function. After enhancing the low-frequency and high-frequency components respectively, an inverse transform reconstruction is performed, and the sub-apertures are stitched together to obtain the enhanced image. Step 2: Set the number of aberration modes to be restored to n, and calculate the centroid offset matrix D of all sub-aperture spots with each aberration mode as input. The dimension is 2m×n. The inverse matrix of D is the mode restoration matrix R generated by the sub-aperture, with the dimension of n×2m. Step 3: Calculate the centroid data x of each sub-spot in the enhanced spot array image using spot localization technology. c y c The slope vector S of the current probe wavefront is obtained by sub-spot offset calculation. The dimension is 2m×1. The restored n-order aberration mode coefficients are obtained by A=R·S, where R is the aberration mode coefficient restoration matrix. Based on the aberration mode and the corresponding coefficients, the wavefront is reconstructed by summing them up.

2. The wavefront restoration algorithm for a Shaker-Hartmann wavefront sensor based on image enhancement according to claim 1, characterized in that: The image enhancement methods described in step 1 include wavelet transform, low-pass and high-pass filtering, and homomorphic filtering image enhancement methods.

3. The wavefront restoration algorithm for a Shaker-Hartmann wavefront sensor based on image enhancement according to claim 1, characterized in that: The spot localization techniques described in step 3 include windowing, weighted centroid method, threshold centroid method, matched filtering method, and registration-based algorithm spot localization method.

4. The wavefront restoration algorithm for a Shaker-Hartmann wavefront sensor based on image enhancement according to claim 1, characterized in that: The aberration modes described in steps 2 and 3 are Zernike aberration mode and Legendre aberration mode.

Images