Target wavefront detection with variable scalability is a technical challenge for adaptive optics applications in object identification, laser atmospheric transmission, laser communication, object imaging, and other fields. In engineering applications within these fields, the phase information of an aberrated wavefront is measured in real time using a correlated Hartmann wavefront sensor. This sensor contributes to excellent wavefront correction in adaptive optics systems, which can lead to imaging quality optimization and beam quality enhancement. As the “eye” of an adaptive optics system, the detection accuracy of the wavefront sensor directly affects the correction capability of the adaptive optics. Hartmann wavefront detection of extended objects often involves the use of the cross-correlation function to estimate the peak value to obtain the sub-aperture offset and then recover the wavefront phase. When the detector is operated, owing to the real-time change in the motion state of the extended object itself and the influence of atmospheric turbulence, the image of the extended object of the sub-aperture object surface often exhibits various types of quality degradation, such as image contrast change, spot drift, and spot dispersion. The flicker phenomenon most often occurs on the edge sub-aperture, which directly leads to the partial loss of the sub-aperture image, thus affecting the calculation of the sub-aperture offset by the relevant algorithm and reducing the accuracy of wavefront detection. In this paper, we report on the sub-aperture and reference images required by an image preprocessing algorithm. After processing, the calculation accuracy of the sub-aperture offset is greatly improved. Under different missing degrees of the image, the preprocessing method exhibits stability, and the wavefront detection accuracy is improved accordingly.

In this study, the speeded-up robust feature (SURF) feature-matching image preprocessing method is used. Before the cross-correlation calculation of the sub-aperture offset, the reference image near the central position of the selected sub-aperture imaging quality and the sub-aperture image under test are used as the feature-matching image pairs. For feature extraction, description, and matching, the pixel unit block with the highest matching degree is selected, that is, the pixel unit block with the shortest Euclidean distance, and the remaining image parts are set to zero and deleted as a new input image pair for cross-correlation calculation. The offset to each sub-aperture is calculated, and the slope matrix of the entire sub-aperture is calculated with the extracted calibrated reference position. Subsequently, the mature Zernike mode method is used to reconstruct the wavefront to be measured, and the wavefront detection process is completed.

The influence of different missing degrees of the sub-aperture image of the extended object on wavefront recovery is analyzed, and a quantitative description of the degree of expansion is provided. The influence of partially missing degrees of the sub-aperture image on wavefront detection accuracy under different expansion degrees, sub-aperture resolutions, and signal-to-noise ratios is investigated, and a detection method based on SURF feature-matching preprocessing is proposed. With an increase in the partial missing degree of the sub-aperture image, the error of the slope extraction result from the cross-correlation algorithm increases, and the system wavefront detection accuracy decreases. Slope extraction after SURF feature-matching preprocessing is significantly improved but limited by the sub-aperture resolution of the detection system; the lower the resolution, the worse the improvement effect. It is also affected by the system noise; when the signal-to-noise ratio is too low, there is a mismatching pair, resulting in a larger error on the slope extraction. Under the same degree of missing image, the greater the degree of expansion, the lower the object detection accuracy. Compared with the case in which the sub-aperture image is less missing, SURF feature-matching preprocessing can be performed after improving the sub-aperture resolution to compensate for the slope extraction error caused by the loss. The wavefront to be measured can then be restored, thereby improving the detection effect of the adaptive optics (AO) system.

In this study, a numerical model of correlated Hartmann wavefront detection for extended objects in the case of partial loss of the sub-aperture image is established. The degree of expansion of an extended object is quantitatively described, and the influence of the degree of sub-aperture image loss on slope extraction and the detection accuracy of the correlated Hartmann wavefront is analyzed. A detection method based on SURF feature-matching preprocessing is proposed. The simulation results show that the slope extraction error caused by partial loss of the sub-aperture image of a reasonable extended object can be reduced by 66%. The peak-to-valley and root mean square values of wavefront restoration residuals are reduced, and the detection accuracy of the adaptive optical wavefront is significantly improved. The results provide a reference for the design of the Hartmann sensor. In the future, experimental research will be conducted using the object characteristic of the non-cooperative variable-extension degree of motion. Moreover, image enhancement technology will be further studied when the sub-aperture imaging part of the target is missing in the case of strong scintillation, to improve the Hartmann wavefront sensing ability.