Solar prominences exhibit intricate fine structures and dynamic evolution that are crucial for understanding solar physics and space weather phenomena. Ground-based large-aperture optical telescopes equipped with adaptive optics (AO) systems can provide high-resolution images of solar prominences. However, a key challenge for solar prominence AO systems is extracting wavefront error information from the faint prominence structures. This study investigates and compares wavefront detection methods based on the cross-correlation coefficient (CC) and absolute difference (AD) algorithms for prominence images captured at the Hα wavelength (6563 ?) to evaluate their feasibility and performance under real observational conditions.

Using a Hartmann-Shack wavefront sensor equipped with a narrow-band Hα filter and a C-Blue One camera, 1500 frames of solar prominence images were collected at a frame rate of 1301 Hz with a spatial resolution of 1 (″)/pixel. The sensor aperture consisted of 9×8 sub-apertures, of which 42 sub-apertures were effective. Image preprocessing involved dark-field and flat-field corrections to mitigate camera non-uniformity and background noise. Wavefront slopes of each sub-aperture were derived through sub-image correlation against a reference sub-image containing solar prominence features. Two correlation-based algorithms, the cross-correlation coefficient (CC) and absolute difference (AD), were implemented to compute the sub-aperture offsets. The extracted slope data over 1500 frames were further analyzed to generate temporal sequences, enabling computation of power spectral density (PSD) and signal-to-noise ratio (SNR) for noise quantification.

The results demonstrate that both CC and AD algorithms can effectively extract the sub-aperture slopes and overall wavefront tilt signals from prominence images. Spatial distributions of sub-aperture offsets derived by both algorithms for a single frame exhibited strong consistency (Figs. 5 and 6), while time sequence of slope measurements showed stable behavior (Figs. 7 and 8). Comparison of overall wavefront tilt sequences further confirmed methodological agreement (Fig. 9). PSD analysis of the extracted signals revealed that the CC algorithm consistently generated lower noise power in the high-frequency regime than the AD algorithm (Figs. 10 and 11). Quantitative SNR evaluation demonstrated an advantage of the CC algorithm, with average values reaching 19.00 dB versus 9.52 dB for the AD algorithm, indicating approximately 8.86 times linear SNR for CC (Table 1, Fig. 12). These findings show that the CC method has superior noise performance for wavefront detection in the context of faint prominence structures.

This study validates the feasibility of correlation-based wavefront detection using real solar prominence images, demonstrating the CC algorithm’s superior noise suppression compared to the AD algorithm. The successful extraction of sub-aperture slopes and overall wavefront tilt signals at the Hα wavelength establishes the CC algorithm’s applicability for solar prominence AO systems. Future work will focus on incorporating higher-order aberration correction to enhance detection accuracy and optimize AO system performance for solar prominence observations.