Wavefront Reconstruction Method Based on Improved U-Net

Laser transmission in the atmosphere is subject to atmospheric turbulence effects that cause wavefront aberrations, resulting in a degradation of beam quality and difficulty in obtaining information. In order to correct the wavefront aberration, it is extremely important to accurately obtain the phase information of beam aberration wavefront for adaptive optics and laser atmospheric transmission system. In this paper, a wavefront reconstruction model based on improved U-Net is established to directly reconstruct the aberrated wavefront by training the aberrated wavefront and the far-field focal plane and out-of-focus plane images, and at the same time, in order to study the relationship between the reconstructed wavefront accuracy and the number of Zernike terms used, the reconstructed wavefronts are represented by the first 50, first 120, first 200, and first 500 Zernike polynomials, respectively. The experimental outcomes illustrate that the wavefront reconstruction precision of the improved U-Net is higher, and the more Zernike polynomial terms used in the reconstructed wavefront, the smaller the coefficient error and the higher the recovered wavefront accuracy. The method can provide strong support for the subsequent wavefront aberration correction, and has certain application value in the fields of turbulence compensation, atmospheric turbulence characterization and adaptive optics.