Curved hologram generation method for speckle noise suppression based on the stochastic gradient descent algorithm

In this paper, a curved hologram generation method with suppressed speckle noise is proposed. In the process of generating the curved hologram, the angle spectrum method is used to calculate the 3D object in layers. By analyzing the loss function relationship between the diffraction image of the curved hologram and the target light field, the loss function is calculated. The phase of the hologram is updated based on the stochastic gradient descent algorithm, thereby obtaining the optimal phase distribution of the curved hologram. When the curved hologram is reproduced at different bending center angles, the reconstructed image of the corresponding angle can be seen. The proposed method has obvious advantages compared with the error diffusion method and Fresnel diffraction algorithm. The feasibility of the proposed method is verified by the experiments.