A Shack-Hartmann based centroid calculation acceleration method

Abstract

The application discloses a Shack-Hartmann-based centroid calculation acceleration method, which determines the effective row pixel range of a sub-aperture by the spot intensity information in the Shack-Hartmann sub-aperture, selects the effective pixels in a single Shack-Hartmann sub-aperture for centroid calculation, avoids waiting for the invalid pixels in the sub-aperture in the image acquisition process, thereby accelerating the centroid calculation speed, and improving the control bandwidth of an adaptive optical system without affecting the wavefront recovery accuracy.

Description

Technical Field

[0001] This invention belongs to the field of image acquisition and signal processing, specifically relating to a centroid calculation acceleration method based on Shaker-Hartmann, which can be used to improve the control bandwidth of adaptive optics systems and expand the system's practical application range. Background Technology

[0002] An adaptive optics system is a system capable of detecting and compensating for wavefront distortion in real time. It consists of three parts: a wavefront detector, a wavefront controller, and a wavefront corrector. Typically, a wavefront processor obtains the wavefront distortion information detected by the wavefront detector and calculates the control voltage of the wavefront corrector to drive a deformable mirror to compensate for the wavefront distortion. The Shaker-Hartmann wavefront sensor (SH-WFS) has advantages such as simple structure, fast measurement speed, and strong adaptability, and is widely used in various fields such as laser transmission, astronomical observation, and medical imaging. This sensor consists of a microlens array and a CCD. Based on the proportional relationship between the centroid displacement of a sub-spot and the average slope of the corresponding sub-aperture wavefront, the local wavefront slope is estimated by calculating the centroid displacement of each sub-spot. Finally, the entire incident wavefront is reconstructed using a corresponding wavefront restoration algorithm based on the sub-aperture slope information.

[0003] The advantages of FPGA-based adaptive optics wavefront processors include good real-time performance and strong parallel processing capabilities. The specific processing procedure of an FPGA-based wavefront processor is as follows: Figure 1 As shown, while achieving pipelined processing of image acquisition, slope calculation, and wavefront restoration, parallel processing of the three stages can also be realized. That is, during the acquisition of the next Hartmann sub-aperture image, the slope calculation and local wavefront restoration of the previous sub-aperture can be performed. With the development of adaptive optics, the number of Hartmann sub-apertures and deformable mirror actuators in adaptive optics systems is constantly increasing. To ensure a high closed-loop control bandwidth, the need for algorithm acceleration and optimization schemes for adaptive optics systems is urgent. Summary of the Invention

[0004] The main problem addressed by this invention is to propose a method for accelerating centroid calculation based on the Shaker-Hartmann method. This method primarily improves the closed-loop bandwidth by reducing the calculation of invalid pixels in the Hartmann sub-aperture. First, the acquired pixel data is denoised. Then, the average value of each row of pixels in the sub-aperture is calculated sequentially. Next, the average values ​​of several consecutive rows are compared. Finally, by determining whether a set condition is met, valid rows of pixels in the sub-aperture are selected for centroid calculation, thus reducing the acquisition and calculation time for invalid rows of pixels.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0006] A Shack-Hartmann centroid calculation acceleration method, which calculates the centroid by screening effective row pixels in a Shack-Hartmann sub-aperture; the method is realized by the following steps:

[0007] Step 1: image acquisition through a Camera-link interface, setting a noise threshold according to statistical rules or experience and carrying out denoising processing;

[0008] Step 2: calculating the Airy spot radius pixel number PNum according to the Hartmann design parameters through the microlens array focal length and the incident wavefront wavelength, and PNum = ┍Num-Mparameter┓, wherein Mparameter is a correction parameter, which is selected by the wavefront distortion degree, ┍┓ is the upward rounding symbol, Num is the final selected Airy spot radius pixel value, and - represents parameter subtraction;

[0009] Step 3: accumulating and averaging the collected sub-aperture row pixels, and then determining the single Shack-Hartmann sub-aperture row effective pixel range;

[0010] Step 4: after determining the effective pixel row, judging whether the sum of the effective pixel row and the Airy spot radius pixel value Num is less than the single Shack-Hartmann sub-aperture row number, if yes, carrying out centroid calculation, otherwise taking the single sub-aperture complete pixel as the effective pixel.

[0011] Further, the denoising processing of step 1 is based on the threshold center method, and the processing flow is as follows:

[0012]

[0013] Wherein, Pixelvalue is the pixel value obtained by image acquisition, newPixelvalue is the pixel value after processing, and threhold is the pixel threshold.

[0014] Further, the detailed process of determining the single Shack-Hartmann sub-aperture row effective pixel range in step 3 is as follows:

[0015] Step 3.1: first, summing the single sub-aperture row pixels, and then averaging each row, and the processing flow is as follows:

[0016]

[0017] Wherein, M is the number of pixels occupied by a single Shack-Hartmann sub-aperture row, I is the pixel gray value, and i is the pixel row number;

[0018] Step 3.2: then, sorting the obtained row averages, if L h ≥L h+1 …≥Lh+Num ,

[0019] h+Num is the effective pixel row in a single sub-aperture, wherein L h is the average value of the hth row of pixels in a single Shack-Hartmann sub-aperture, L h+1 is the average value of the (h+1)th row of pixels in a single Shack-Hartmann sub-aperture, L h+Num is the average value of the (h+Num)th row of pixels in a single Shack-Hartmann sub-aperture.

[0020] Further, the processing procedure of the centroid calculation in step 4 is as follows:

[0021]

[0022] wherein X and Y represent the calculated centroid coordinates; x and y are the pixel coordinates of a single Shack-Hartmann wavefront sensor, and I is the corresponding light intensity gray value.

[0023] Compared with the prior art, the present application has the following advantages:

[0024] The centroid calculation on the effective row pixels in the sub-aperture can effectively speed up the centroid calculation speed without affecting the detection precision, and improve the closed-loop bandwidth of the adaptive optical system. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is the centroid calculation flow chart in the embodiment one of the present application.

[0026] Figure 2 is the timing diagram of the Hartmann single sub-aperture averaging module processing in the embodiment one of the present application.

[0027] Figure 3 is the detailed signal processing process diagram of the coordinate x and gray value multiplication module, the coordinate y and gray value multiplication module, and the pixel gray value summation module in the embodiment one of the present application.

[0028] Figure 4 is the Hartmann image with white noise in the embodiment one of the present application.

[0029] Figure 5 is the Hartmann image after the centroid acceleration scheme processing in the embodiment one of the present application.

[0030] Figure 6 is the multiple-experiment wavefront recovery RMS comparison diagram in the embodiment one of the present application. DETAILED DESCRIPTION

[0031] In order to make the purpose, circuit design scheme and advantages of the present application more clear and intuitive, the present application is further described in detail below in combination with the drawings and the specific embodiment one.

[0032] The present application provides a Shack-Hartmann centroid calculation acceleration method, and specific embodiments are shown in the accompanying drawings Figure 1 The data processing flow is as follows:

[0033] 1. Image acquisition is performed, and then denoising is performed.

[0034] 2. Parallel entry into the coordinate x and gray value multiplication module, the coordinate y and gray value multiplication module, the pixel gray value summation module, and the sub-aperture mean value processing module.

[0035] 3. After the coordinate and gray value multiplication is completed in the x and gray value multiplication module and the y and gray value multiplication module, the summation is performed in the double-channel accumulator.

[0036] 4. In the mean value processing module, the multiplication mainly completes the summation of the single Hartmann sub-aperture row pixels and the determination of the mean value of each row, and then the effective pixels of the single Hartmann sub-aperture row are determined.

[0037] 5. After the effective pixel row is determined, it is judged whether the number of effective rows and the number of pixels occupied by the Airy disk radius is less than the number of single Hartmann sub-aperture rows, and if so, the centroid calculation is performed, otherwise the complete pixels of the single sub-aperture are taken as the effective pixels.

[0038] The mean value module mainly realizes the mean value of the row pixels, and then enters the sorting module. If the requirements are met, the centroid calculation can be completed in advance, and the detailed signal process is shown in Figure 2 . The Sclk is the camera sampling clock frequency, the Line_Sum is the row pixel summation signal, the Xclk is the Sclk three times frequency as the signal processing frequency, the Line_div is the mean value processing signal, the Line_div_delay is the mean value processing signal delay signal, the Line_flag is the judgment flag bit, and the cnt_Num is the effective row count signal. When the row start signal is received, the Line_Sum starts to accumulate the row signal. When the row end signal is received, the row pixel summation signal is mean value processed, and then it is judged whether the Line_div is greater than the Line_div_delay. When the condition is established, the Line_flag level is converted to high level. When the Line_flag is high level and the Line_div is greater than the Line_div_delay, the cnt_Num is accumulated. Secondly, if the next row still meets the condition that the Line_div is greater than the Line_div_delay, the cnt_Num is accumulated, otherwise the Line_flag is converted to low level, and the cnt_Num is cleared. When the cnt_Num is equal to the Num, the Flag is converted to high level.

[0039] The coordinate x and gray value multiplication module, the coordinate y and gray value multiplication module, and the pixel gray value summation module are shown in detail in the signal processing process as shown in Figure 3 Cnt_colum is the column counter, Cnt_row is the row counter, div_cal is the division signal, (Line+Num-1) is the effective pixel row, and other signals are the same as those in the mean value processing module. When the Flag signal is high, the division operation is started, and when the falling edge of div_cal is detected, the Flag flag signal is turned to low. When (Line+Num-1) is greater than M, the Flag flag signal is low, and the centroid is not accelerated.

[0040] The distorted wavefront is generated by means of the statistical characteristics of atmospheric Kolmogorov turbulence, and image data is obtained in the Hartmann system. The Hartmann sub-aperture is 10x10, the sub-aperture pixel is 24x24, the sub-aperture shape is square, and the sub-aperture arrangement shape is square. The number of pixels occupied by the Airy disk radius is 3, and the Mparameter correction parameter is 2. The random wave surface generated by the Airy disk is divided, sampled and focused on the CCD through the microlens array to form a spot array image. On the basis of the spot array formed on the CCD, Gaussian white noise with a signal-to-noise ratio of 10% is added, and the obtained spot image is as shown in Figure 4 The obtained CCD image array data with noise is obtained by hardware function simulation to obtain centroid data and effective pixel cutoff row, and then the slope is calculated and the wavefront is restored. Figure 5 is the CCD spot array image after the centroid acceleration method, wherein the white dashed line in the figure represents the effective pixel cutoff row. It can be seen from Figure 5 that the centroid calculation speed is improved to different degrees for each Hartmann sub-aperture. The centroid acceleration method is tested for many times, and the slopes obtained by the centroid acceleration method in the case of noise and the traditional centroid calculation method in the case of denoising are compared in terms of wavefront restoration residual error as shown in Figure 6 The dashed line in the figure represents the wavefront restoration residual error in the case of noise, the asterisk (*) in the figure represents the traditional restoration noise in the case of denoising, and the purple color in the figure represents the wavefront restoration residual error of the centroid acceleration method in the case of denoising. It can be seen that the centroid accuracy of the centroid acceleration method in the case of denoising and the traditional centroid calculation method in the case of denoising is consistent, so the method can improve the centroid calculation speed without affecting the wavefront restoration accuracy, thereby providing feasibility for improving the closed-loop control bandwidth of the adaptive optical system.

[0041] The present application is not limited to the above-mentioned specific embodiments, and the above-mentioned embodiments are only illustrative, but not restrictive. A person of ordinary skill in the art can make many forms without departing from the purpose of the present application and the scope protected by the claims under the inspiration of the present application, and these all belong to the protection of the present application. The part of the present application not described in detail belongs to the known technology of the person of ordinary skill in the art.

Claims

1. A method for accelerating centroid calculation based on Shaker-Hartmann, characterized in that: The method calculates the centroid by filtering the effective pixel rows in the Shack-Hartmann sub-aperture; this method is specifically implemented through the following steps: Step 1: Acquire images via the Camera-link interface, set a noise threshold based on statistical patterns or experience, and perform noise reduction processing; Step 2: Based on the design parameters of the Hartmann microlens array focal length and the incident wavefront wavelength, calculate the number of pixels occupied by the final selected Airy disk radius as Num, and Num = ┍PNum - Mparameter┓, where Mparameter is the correction parameter, which is selected according to the degree of wavefront distortion, ┍ ┓ is the rounding up sign, PNum is the value of the number of pixels occupied by the selected Airy disk radius before correction, and - indicates the subtraction of parameters; Step 3: Accumulate and average the collected sub-aperture row pixels, and then determine the effective pixel row range of a single Shaker-Hartmann sub-aperture; Step 4: After determining the effective pixel row, check whether the sum of the effective pixel row and the number of pixels Num occupied by the final selected Airy disk radius is less than the number of rows of a single Shaker-Hartmann sub-aperture. If it is true, perform centroid calculation; otherwise, take the complete pixel of a single sub-aperture as the effective pixel.

2. The centroid calculation acceleration method based on Shaker-Hartmann according to claim 1, characterized in that: The denoising process described in step 1 is based on the threshold centroid method, and its processing flow is as follows: Where Pixelvalue is the pixel value obtained from image acquisition, newPixelvalue is the processed pixel value, and threshold is the pixel threshold.

3. The centroid calculation acceleration method based on Shaker-Hartmann according to claim 1, characterized in that: The detailed process for determining the effective pixel range of a single Shaker-Hartmann sub-aperture row in step 3 is as follows: Step 3.1: First, sum the pixels in each sub-aperture row, then calculate the mean of each row. The processing flow is as follows: Where M is the number of pixels occupied by a single Shaker-Hartmann sub-aperture row, I is the pixel gray value, and i is the number of pixel rows; Step 3.2: Then sort the calculated row means. If the following conditions are met... , Then h+Num represents the effective pixel row in a single sub-aperture, where It is the average pixel value of the h-th row of a single Shaker-Hartmann sub-aperture. It is the average pixel value of the (h+1)th row of a single Shaker-Hartmann sub-aperture. It is the average pixel value of the h+Num row of a single Shaker-Hartmann sub-aperture.

4. The centroid calculation acceleration method based on Shaker-Hartmann according to claim 3, characterized in that: The process of centroid calculation described in step 4 is as follows: Where X and Y represent the calculated centroid coordinates; x and y are the pixel coordinates of a single Shaker-Hartmann wavefront sensor, and I is the corresponding light intensity grayscale value.

Images