Holographic video rapid optimization method based on clustering algorithm

Abstract

The invention provides a fast holographic video optimization method based on a clustering algorithm for solving the problem that the calculation speed of holographic video optimization is low. According to the method, the calculation complexity in holographic video optimization is remarkably reduced by introducing the clustering algorithm. The optimized hologram of the center of the cluster is obtained firstly and serves as the initial hologram of other frames in the cluster, the optimization speed is increased, and the quality of the optimization result is guaranteed. According to the method, while holographic video optimization calculation is accelerated, high reconstruction quality is kept, and the method has a wide application prospect.

Description

Technical Field

[0001] The invention relates to a holographic display technology, in particular to a method for generating a hologram. Background Art

[0002] Holographic display technology is an advanced display technology that can present three-dimensional images and videos, with extremely high visual immersion and authenticity. In a holographic display system, the generation of images and videos requires a complex calculation process, especially in dynamic scenes, the optimization of holographic videos is particularly challenging. At present, the stochastic gradient descent algorithm is widely used in the optimization of holographic images because of its high efficiency in processing image details and optimizing errors. However, for the optimization of holographic videos, the computational cost of the stochastic gradient descent method is high, because each frame of the image needs to be optimized separately, which obviously has a large performance bottleneck for real-time display and long-term playback scenarios. Therefore, how to improve the calculation speed of holographic video optimization and reduce the time and resources required for optimization has become a key issue that needs to be solved in this field. Summary of the invention

[0003] Aiming at the problem of slow calculation speed of holographic video optimization, the present invention provides a fast optimization method for holographic video based on clustering algorithm, which is characterized by the following method: ① clustering each frame in the holographic video by using clustering algorithm, and determining the center of each cluster cluster according to the clustering result; ② initializing a random phase θ0 for the center of each cluster cluster, and generating an initial hologram us0=exp(i*θ0) based on the phase; ③ reconstructing the hologram by using the diffraction model to obtain the reproduced light field ud; ④ substituting the amplitude Ampr of the reproduced light field and the target image Ampt into the loss function for calculation to obtain the loss value L, which is expressed as L=LOSS(Ampr, Ampt); ⑤ According to the loss value, calculate its gradient with respect to the hologram phase; ⑥ Search along the direction of the gradient descent of the loss value with a step size step to update the phase; ⑦ Repeat steps ③-⑥ to continuously optimize and update the hologram phase. When the set number of optimizations N1 is met or the expected reconstruction quality is achieved, the iteration is terminated, and finally the optimized hologram us1=exp(i*θ1) of the center of the cluster is obtained; ⑧ Use the optimized hologram of the center of the cluster as the initial hologram of all frames in the cluster; ⑨ For each frame of the hologram, repeat steps ③-⑥ for optimization until the set number of optimizations N2 is reached or the expected reconstruction quality is achieved, thereby obtaining the final hologram of each frame.

[0004] The present invention significantly reduces the computational complexity of holographic video optimization by introducing a clustering algorithm. By first optimizing the central frame of the cluster and then using it as the initial hologram of other frames in the cluster, not only the optimization speed is improved, but also the quality of the optimization result is guaranteed. This method accelerates the optimization calculation of holographic video while maintaining a high reconstruction quality, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Attached Figure 1 It is a schematic diagram of the process of the present invention.

[0006] Attached Figure 2 The figure compares the simulation results and optical results of the method of the present invention and the original invention method under the same number of optimizations.

[0007] Attached Figure 3 The figure compares the number of optimization times required to optimize the holographic video to the same quality between the method of the present invention and the original invention method. DETAILED DESCRIPTION

[0008] A typical embodiment of a solution to the problem of cylindrical computer-generated holographic high-order diffraction images of the present invention is described in detail below, and the present invention is further described in detail. It is necessary to point out here that the following embodiments are only used to further illustrate the present invention and cannot be understood as limiting the scope of protection of the present invention. Personnel skilled in the art in this field may make some non-essential improvements and adjustments to the present invention based on the above content of the present invention, which still fall within the scope of protection of the present invention.

[0009] The specific implementation of the method of the present invention is as follows:

[0010] Aiming at the problem of slow calculation speed of holographic video optimization, the present invention provides a fast optimization method for holographic video based on clustering algorithm, which is characterized by the following method: ① clustering each frame in the holographic video by using clustering algorithm, and determining the center of each cluster cluster according to the clustering result; ② initializing a random phase θ0 for the center of each cluster cluster, and generating an initial hologram us0=exp(i*θ0) based on the phase; ③ reconstructing the hologram by using the diffraction model to obtain the reproduced light field ud; ④ substituting the amplitude Ampr of the reproduced light field and the target image Ampt into the loss function for calculation to obtain the loss value L, which is expressed as L=LOSS(Ampr, Ampt); ⑤ According to the loss value, calculate its gradient with respect to the hologram phase; ⑥ Search along the direction of the gradient descent of the loss value with a step size step to update the phase; ⑦ Repeat steps ③-⑥ to continuously optimize and update the hologram phase. When the set number of optimizations N1 is met or the expected reconstruction quality is achieved, the iteration is terminated, and finally the optimized hologram us1=exp(i*θ1) of the center of the cluster is obtained; ⑧ Use the optimized hologram of the center of the cluster as the initial hologram of all frames in the cluster; ⑨ For each frame of the hologram, repeat steps ③-⑥ for optimization until the set number of optimizations N2 is reached or the expected reconstruction quality is achieved, thereby obtaining the final hologram of each frame.

[0011] In an example of the present invention, a fast optimization method for holographic video based on a clustering algorithm is provided, and the present invention utilizes a K-means clustering algorithm as a clustering algorithm.

[0012] In an example of the present invention, a fast optimization method for holographic video based on a clustering algorithm is provided. The present invention uses an angular spectrum diffraction algorithm as a diffraction model, which is expressed as P(us)=IFT[FT(us)*T], where FT and IFT are Fourier transform and inverse Fourier transform, respectively. T is a transfer function, which is expressed as exp{j2πd / λ*sqrt[1-λ^2*(fx^2+fy^2)]}, where d is the diffraction distance, λ is the wavelength, fx and fy are frequency domain coordinates, and sqrt is the square root.

[0013] In an example of the present invention, a solution to the problem of cylindrical computer-generated holographic high-order diffraction images is provided. The present invention uses a mean square error loss function as a loss function.

[0014] The content of the present invention is further explained below in conjunction with embodiments and drawings:

[0015] During the hologram calculation and generation process, the resolution of the hologram is 1080×1920, the diffraction distance d is 25cm, the wavelength λ is 532nm, the step size step is 0.02, and the optimization number N1 is 100. Figure 2The simulation results and optical results of the method of the present invention and the original invention method under the same number of optimization times are compared. From the results, it can be seen that under the same number of optimization times, the simulation results and optical results of the method of the present invention are better. Figure 3 The number of optimization times required for optimizing a holographic video to the same quality between the method of the present invention and the original invention method is compared. From the results, the method of the present invention requires fewer optimization times, which effectively improves the calculation speed of holographic video optimization and reduces the time and resources required for optimization.

Claims

1. A fast optimization method for holographic video based on clustering algorithm, characterized by The method is as follows: ① Cluster each frame in the holographic video using a clustering algorithm, and determine the center of each cluster based on the clustering results; ② Initialize a random phase θ0 for the center of each cluster, and generate an initial hologram us0=exp(i*θ0) based on the phase; ③ Use the diffraction model to reconstruct the hologram and obtain the reproduced light field ud; ④ Substitute the amplitude Ampr of the reproduced light field and the target image Ampt into the loss function for calculation to obtain the loss value L, which is expressed as L=LOSS(Ampr, Ampt); ⑤ According to the loss value, calculate its gradient with respect to the hologram phase; ⑥ Search along the direction of the gradient descent of the loss value with a step size step to update the phase; ⑦ Repeat steps ③-⑥ to continuously optimize and update the hologram phase. When the set number of optimizations N1 is met or the expected reconstruction quality is achieved, the iteration is terminated, and finally the optimized hologram us1=exp(i*θ1) of the center of the cluster is obtained; ⑧ Use the optimized hologram of the center of the cluster as the initial hologram of all frames in the cluster; ⑨ For each frame of the hologram, repeat steps ③-⑥ for optimization until the set number of optimizations N2 is reached or the expected reconstruction quality is achieved, thereby obtaining the final hologram of each frame.

2. The holographic video fast optimization method based on clustering algorithm according to claim 1 is characterized in that: The clustering algorithm described in step ① may be a K-means clustering algorithm, a hierarchical clustering algorithm, a mean shift clustering algorithm, or other clustering algorithms.

3. The holographic video fast optimization method based on clustering algorithm according to claim 1 is characterized in that: The diffraction model described in step ③ may be an angular spectrum diffraction model, a Rayleigh-Sommerfeld diffraction model, a Kirchhoff diffraction model, or other diffraction models.

4. The holographic video fast optimization method based on clustering algorithm according to claim 1, characterized in that: The loss function described in step ④ may be mean square error, structural similarity, or other loss functions.

Images