Naked eye 3D data communication method

Abstract

The invention relates to the technical field of naked-eye 3D display, and discloses a naked-eye 3D data communication method, which comprises the following steps of S1, acquiring light field data through multi-view light field acquisition equipment, performing normalization processing on the light field data, and performing noise removal on the light field data by adopting a filtering method; and S2, performing variational light field optimization on the preprocessed light field data, and by constructing a light field gradient regularization model, representing the light field data by using an orthogonal basis function, so that the dimension of the light field data is reduced to low-dimensional light field representation. An autoregressive moving average model is utilized to predict a bandwidth state, a self-adaptive bandwidth regulation and control strategy is combined, and a data transmission priority is dynamically adjusted according to a bandwidth threshold value, so that compared with fixed code rate transmission and simple bandwidth monitoring and adjustment in the prior art, the bandwidth fluctuation trend can be predicted in advance, the problem of data loss caused by sudden bandwidth decline can be reduced, and the data transmission efficiency can be improved. And naked eye 3D data communication is stable.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of naked eye 3D display, in particular to a naked eye 3D data communication method. BACKGROUND

[0002] The naked eye 3D data communication technology aims to provide users with a three-dimensional visual experience without wearing any auxiliary equipment. To achieve this goal, efficient light field data acquisition, processing, encoding, transmission and recovery technology is needed. Although existing methods provide solutions to existing problems, there are still various limitations in practical applications, affecting the effect and stability of naked eye 3D communication.

[0003] The current mainstream light field data dimension reduction method mostly uses principal component analysis and singular value decomposition, and the main idea is to find a low-dimensional representation of the data through linear transformation. The global feature extraction of the method makes the ability to maintain local details weak, especially the edge information and high-frequency features in the light field data are easily lost, resulting in the decline of the quality of the reconstructed light field after dimension reduction. In addition, principal component analysis and singular value decomposition usually rely on large-scale matrix decomposition, which has high computational complexity and cannot run efficiently in real-time naked eye 3D communication applications.

[0004] Naked eye 3D data communication is very sensitive to network bandwidth fluctuations, and existing data transmission strategies are mostly based on fixed code rate transmission and simple bandwidth monitoring adjustment. Fixed code rate transmission can ensure data quality, but when the bandwidth fluctuates greatly, it is easy to cause data loss and lag, affecting the viewing experience. The adjustment method based on simple bandwidth monitoring is usually a passive strategy and cannot predict the trend of bandwidth changes in advance, resulting in the inability to adjust the transmission strategy in time when the bandwidth drops suddenly, and the picture quality decreases or even the transmission is interrupted.

[0005] Due to data loss and bandwidth limitation in the transmission process, the receiving end needs to recover the light field data. The existing recovery method mainly uses simple interpolation and mean compensation, which has low computational complexity, but lacks constraints on the structure of the light field data in the recovery process, resulting in deviations between the recovery results and the original data.

[0006] Therefore, the naked eye 3D data communication method provided by the person skilled in the art solves the above-mentioned problems. SUMMARY

[0007] In view of the shortcomings of the prior art, the naked eye 3D data communication method is provided to solve the problems in the background art.

[0008] To achieve the above purpose, the naked eye 3D data communication method is implemented by the following technical scheme: a naked eye 3D data communication method, comprising the following steps:

[0009] Step S1, obtaining light field data by a multi-view light field acquisition device, and performing normalization processing on the light field data, and then removing noise from the light field data by a filtering method;

[0010] Step S2, performing variational light field optimization on the preprocessed light field data, constructing a light field gradient regularization model, representing the light field data by using an orthogonal basis function, and reducing the light field data to a low-dimensional light field representation;

[0011] Step S3, calculating the information distribution of the low-dimensional light field data based on information entropy, dividing the data levels according to the level of information entropy, obtaining the base layer data and the enhancement layer data, and performing multi-scale transformation on the light field data to decompose the light field data into a low-frequency light field component and a plurality of sets of high-frequency light field components of different scales;

[0012] Step S4, during transmission of the light field data, detecting the network bandwidth state in real time, selecting to transmit the base layer data or the combination of the base layer data and the enhancement layer data according to the network bandwidth state, and estimating and compensating for the lost enhancement layer data at the receiving end;

[0013] Step S5, obtaining the transmitted light field data at the receiving end, recovering the untransmitted enhancement layer data or the data lost during transmission based on a variational regularization method, so that the finally received light field data can be used for naked-eye 3D display.

[0014] Preferably, in the step S1, the low-dimensional light field representation is used for subsequent data encoding;

[0015] The normalization processing of the light field data in the step S1 adopts a min-max normalization method, which maps the pixel value of the light field data to a preset range, and the calculation formula is as follows:

[0016] ,

[0017] wherein, is the normalized light field data, is the original light field data, and is the minimum value and the maximum value in the light field data,

[0018] is the spatial coordinate of the light field data, representing the two-dimensional pixel position in the image,

[0019] is the angle coordinate of the light field data, representing the direction information of the light ray in the light field.

[0020] 3. The naked-eye 3D data communication method according to claim 2, wherein the noise removal in the step S1 adopts a Gaussian filtering method, and the mathematical expression is as follows:

[0021] ,

[0022] in, The smoothed light field data, The standard deviation is Two-dimensional Gaussian kernel, The normalized light field data,

[0023] These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image.

[0024] These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

[0025] Preferably, the light field gradient regularization model in step S2 is optimized for dimensionality reduction using the second-order Poisson equation, and its mathematical expression is as follows:

[0026] ,

[0027] in, For the optimized light field data, To observe light field data, As a regularization factor, The Laplacian operator for light field data,

[0028] These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image.

[0029] These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

[0030] Preferably, in step S3, the base layer data contains the main optical field information, and the enhancement layer data contains high-frequency detail information of the optical field;

[0031] In step S4, if the network bandwidth meets the conditions for transmitting enhancement layer data, the base layer data and enhancement layer data are transmitted; if the network bandwidth cannot meet the conditions for enhancing layer data transmission, the base layer data is transmitted.

[0032] Preferably, the information entropy calculation in step S3 uses the Shannon entropy formula, the mathematical expression of which is as follows:

[0033] ,

[0034] in, The information entropy of the light field data, Let be the probability distribution of the i-th gray level in the light field data. The total number of gray levels in the light field data;

[0035] The multi-scale transformation of the light field data in step S3 employs the Laplace pyramid decomposition method, and the mathematical expression is as follows:

[0036] ,

[0037] in, For low-frequency light field components, Let S be the high-frequency optical field component at the s-th scale, where S is the scale level. This is the original light field data.

[0038] These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image.

[0039] These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

[0040] Preferably, the real-time bandwidth status detection in step S4 uses an autoregressive moving average model for prediction, and the mathematical expression is as follows:

[0041] ,

[0042] in, This is the bandwidth prediction value at the current moment. This is the bandwidth measurement value at the current moment. As a smoothing factor, for The bandwidth prediction value at any given time.

[0043] Preferably, the data transmission priority control in step S4 adopts an adaptive bandwidth regulation strategy, and the decision rules are as follows:

[0044] like > At the same time, it transmits a combination of base layer data and enhancement layer data;

[0045] like < At the same time, basic layer data is transmitted, and data compensation is performed at the receiving end.

[0046] in, The set bandwidth threshold.

[0047] Preferably, the data error recovery in step S5 employs a variational regularization method, the mathematical expression of which is as follows:

[0048] ,

[0049] in, For the recovered light field data, The sampling matrix of the light field data, For the received light field data, For regularization parameters, The gradient of the light field data.

[0050] Preferably, the error recovery quality evaluation in step S5 is assessed using peak signal-to-noise ratio and structural similarity index, calculated as follows:

[0051] ,

[0052] ,

[0053] in, Used to evaluate the signal-to-noise ratio between the recovered data and the original data. Mean square error, The maximum pixel value. Used to assess the structural similarity between the recovered data and the original data. and The mean of the dataset. and For variance, For covariance, and It is a stabilizing factor.

[0054] This invention provides a naked-eye 3D data communication method. It has the following beneficial effects:

[0055] 1. This invention employs a second-order Poisson equation-based optical field gradient regularization model to optimize the dimensionality reduction of optical field data. While preserving spatial information, it reduces data redundancy and improves storage and transmission efficiency. Compared to existing technologies that rely on principal component analysis and singular value decomposition for dimensionality reduction, this invention can adaptively preserve the edge details of the optical field during the dimensionality reduction process, avoid information loss, and significantly improve the reconstruction quality of the dimensionality-reduced optical field data.

[0056] 2. This invention uses an autoregressive moving average model to predict bandwidth status and combines it with an adaptive bandwidth control strategy to dynamically adjust data transmission priority based on bandwidth thresholds. Compared with the fixed bit rate transmission and simple bandwidth monitoring and adjustment in the prior art, it can predict bandwidth fluctuation trends in advance, reduce data loss caused by sudden bandwidth drops, and make naked-eye 3D data communication stable.

[0057] 3. This invention uses variational regularization to recover lost data at the receiving end. By optimizing the error minimization target, it accurately compensates for the untransmitted light field data. Compared with the recovery methods using simple interpolation and mean compensation in the prior art, it can adaptively adjust the recovery weight, greatly reduce the impact of missing data on 3D display quality, improve the integrity of the final light field data, and make naked-eye 3D display clear. Attached Figure Description

[0058] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0059] To enable those skilled in the art to understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0060] The present invention will now be described in detail with reference to the accompanying drawings:

[0061] Example:

[0062] Please see the appendix Figure 1 This invention provides a naked-eye 3D data communication method, comprising the following steps:

[0063] Step S1: Acquisition and preprocessing of light field data

[0064] 1. This invention employs a multi-view light field acquisition device to obtain light field data. This device consists of multiple camera arrays, each capturing the scene from different angles, ultimately obtaining light field data containing both spatial and angular information. Unlike traditional single-view and limited-view acquisition schemes, this invention's multi-view light field acquisition method can completely capture parallax information, improving the depth perception of naked-eye 3D displays.

[0065] 2. The acquired light field data pixel values ​​have different dynamic ranges. To unify the data scale, this invention performs minimum-maximum normalization on the light field data. The normalization formula is as follows:

[0066] ,

[0067] in, The normalized light field data, This is the original light field data. and These represent the minimum and maximum values ​​in the light field data.

[0068] Normalized light field data can eliminate brightness differences and improve the robustness of subsequent processing.

[0069] 3. Due to sensor noise during the light field acquisition process, this invention uses Gaussian filtering for noise removal, the mathematical expression of which is as follows:

[0070] ,

[0071] in, The smoothed light field data, The standard deviation is Two-dimensional Gaussian kernel, The normalized light field data,

[0072] These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image.

[0073] These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

[0074] Compared to traditional mean filtering, Gaussian filtering can preserve the edge information of the light field while removing noise, thus avoiding blurring.

[0075] In naked-eye 3D data communication, the quality of light field data directly affects the final 3D visual experience. Therefore, high-quality light field data acquisition and preprocessing are crucial. The core task of step S1 is to acquire high-quality light field data, eliminate noise, and unify data scale, laying the foundation for subsequent data optimization, encoding, and transmission.

[0076] Traditional 3D imaging typically relies on single-view or limited-view techniques, resulting in insufficient parallax information and affecting the 3D effect. This invention employs a multi-view light field acquisition device to ensure scene information is obtained from multiple angles, achieving the following advantages:

[0077] Multi-view data can accurately describe the depth information of a scene, improving the realism of 3D displays.

[0078] Traditional methods can lead to the loss of information from a particular perspective, while multi-view acquisition can provide a natural naked-eye 3D experience.

[0079] Multi-view light field data contains richer angular information, making 3D images clearer.

[0080] Directly processing the raw data increases computational complexity and can even affect the final 3D imaging quality. This invention employs a minimum-maximum normalization method to ensure that the pixel values ​​of all light field data are mapped to the same range, offering the following advantages:

[0081] After normalization, the data is evenly distributed, avoiding the impact of uneven brightness on the 3D imaging effect.

[0082] After standardizing the data scale, subsequent dimensionality reduction, encoding, and restoration operations become more efficient, reducing the accumulation of computational errors.

[0083] Normalization processing adapts data to the display standards of different devices, improving the compatibility of naked-eye 3D technology across different hardware platforms.

[0084] During the light field data acquisition process, the raw data contains noise and photosensitivity due to sensor noise and changes in illumination. Undenoised light field data is prone to distortion during subsequent transmission and reconstruction, affecting the final 3D display effect. This invention uses Gaussian filtering for noise removal, which has the following advantages:

[0085] Gaussian filtering can smooth data, eliminate noise caused by sensors and the environment, and improve data quality.

[0086] The denoised light field data is stable, providing clean input data for subsequent dimensionality reduction, encoding and restoration, and reducing the possibility of error propagation.

[0087] Step S2: Dimensionality reduction and optimization of light field data

[0088] 1. This invention employs an optical field gradient regularization model for dimensionality reduction, constraining the optical field gradient using a second-order Poisson equation to preserve key structural information in the low-dimensional representation. The optimization objective is as follows:

[0089] ,

[0090] in, For the optimized light field data, To observe light field data, As a regularization factor, The Laplacian operator for light field data,

[0091] These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image.

[0092] These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

[0093] Compared to principal component analysis and singular value decomposition dimensionality reduction methods, this scheme can adaptively preserve the light field edges and high-frequency information, avoid image blurring after dimensionality reduction, and improve reconstruction quality.

[0094] Light field data typically possesses high dimensionality, containing rich spatial and angular information. However, the computational demands of high-dimensional data are substantial during data transmission and storage, and direct transmission leads to high bandwidth consumption and decreased processing efficiency. Therefore, this invention introduces a light field gradient regularization dimensionality reduction method in step S2. This method optimizes and reduces the dimensionality of the light field data using a second-order Poisson equation, thereby reducing the data dimensionality while preserving key structural information and improving the quality of data transmission and reconstruction.

[0095] 1. Light field data typically has high dimensionality. Without dimensionality reduction optimization, data storage and transmission will face the following problems:

[0096] High-dimensional light field data contains a large amount of redundant information, resulting in high storage costs and inefficient management.

[0097] Naked-eye 3D displays require real-time transmission of large-scale light field data. High-dimensional data can cause transmission delays, affecting the user experience.

[0098] Directly processing high-dimensional data will significantly increase computational overhead, leading to a decrease in system response speed.

[0099] 2. During dimensionality reduction, traditional methods such as principal component analysis and singular value decomposition can reduce data dimensionality, but they also introduce the following problems:

[0100] Losing important high-frequency details affects the three-dimensionality of the light field data;

[0101] It is not sensitive to edge structures, resulting in blurry and distorted naked-eye 3D images;

[0102] The reduced-dimensionality data cannot be reconstructed, affecting subsequent recovery and enhancement processing.

[0103] In comparison, the optical field gradient regularization dimensionality reduction method of the present invention has the following advantages:

[0104] Adaptively preserves edge and high-frequency information, reducing image blurring after dimensionality reduction;

[0105] Optimize the changes in the light field gradient to make it easier to recover the original structure during reconstruction;

[0106] Compared to traditional principal component analysis and singular value decomposition methods, the data quality after dimensionality reduction is high, which helps to improve the realism of naked-eye 3D displays.

[0107] The light field data is regularized by the second-order Poisson equation to smooth its gradient changes, ensuring that the spatial structure characteristics of the light field are still preserved after dimensionality reduction, so as to achieve high-quality naked-eye 3D reconstruction in the subsequent transmission and recovery process.

[0108] 3. Naked-eye 3D technology needs to ensure smooth and clear visual effects under different devices and bandwidth conditions. Therefore, dimensionality reduction of light field data needs to reduce the amount of data and enhance its adaptability. The dimensionality reduction optimization strategy of this invention can:

[0109] It adapts to naked-eye 3D display devices with different resolutions, ensuring data compatibility across different devices; it reduces the impact of bandwidth fluctuations on display effects, guaranteeing basic 3D display quality even in unstable network environments; and it optimizes data encoding efficiency, making compressed data easier to store and transmit, thereby improving the real-time performance and smoothness of naked-eye 3D video.

[0110] Step S3: Layering and Multi-scale Transformation of Light Field Data

[0111] 1. This invention uses Shannon entropy to calculate the information distribution of light field data, as shown in the following formula:

[0112] ,

[0113] in, The information entropy of the light field data, Let be the probability distribution of the i-th gray level in the light field data. This represents the total number of gray levels in the light field data.

[0114] Based on the level of information entropy, the light field data is divided into:

[0115] Base layer data (low-entropy region, containing main light field information);

[0116] Enhanced layer data (high-entropy regions, including edges and high-frequency details).

[0117] 2. This invention employs Laplace's pyramid decomposition for multi-scale transformation, as shown in the following formula:

[0118] ,

[0119] in, For the high-frequency light field component of the s-th layer, This is the original light field data. It is a Gaussian filter.

[0120] This method can effectively separate low-frequency and high-frequency information, enabling efficient data transmission.

[0121] In glasses-free 3D data communication, the hierarchical structure and information density of light field data have a significant impact on transmission efficiency and display quality. In step S3 of this invention, an information entropy calculation combined with Laplace pyramid decomposition is used to perform layered and multi-scale transformations on the light field data, improving data encoding efficiency, reducing bandwidth consumption, and enhancing the clarity and depth of glasses-free 3D displays.

[0122] 1. Different regions in a light field data have different information densities, for example:

[0123] The main area contains large-scale information about the scene, and the changes are relatively gradual, which has a significant impact on visual perception.

[0124] Edge and detail areas contain high-frequency information of the light field, which is crucial for clarity and three-dimensionality.

[0125] Based on the information entropy calculation results, the light field data is adaptively divided into:

[0126] The base layer data contains the main optical field information and is the primary data to be transmitted, so transmission stability is given priority.

[0127] The enhancement layer data contains details of the optical field and high-frequency information, and the transmission strategy is dynamically adjusted according to the bandwidth.

[0128] The benefits are:

[0129] The base layer ensures basic 3D effects, while the enhancement layer can supplement detailed information and improve display quality when bandwidth allows.

[0130] It prioritizes the transmission of low-entropy data, which can maintain basic 3D visual effects when bandwidth is limited.

[0131] Enhanced layer data can improve edge sharpness and texture clarity, enhancing the 3D stereoscopic effect.

[0132] 2. Light field data contains spatial information and multi-scale features. If this multi-scale information is not properly handled during data compression, transmission, and reconstruction, it can lead to:

[0133] The loss of low-frequency information results in a blurry image;

[0134] Damage to high-frequency information affects the clarity and stereoscopic effect of naked-eye 3D.

[0135] Therefore, this invention employs Laplace's pyramid decomposition for multi-scale transformation. The function of Laplace's pyramid is as follows:

[0136] The base layer data mainly contains low-frequency components, while the enhancement layer data contains high-frequency details, enabling more efficient encoding and transmission.

[0137] The high-frequency portion enables efficient encoding, reduces data redundancy, and improves bandwidth utilization.

[0138] During the reconstruction process, low-frequency information is restored first, and high-frequency details are gradually superimposed to avoid blurring and distortion.

[0139] The benefits are:

[0140] Necessary high-frequency information is additionally encoded to avoid transmitting unnecessary data.

[0141] Dynamically adjust the transmission strategy for high-frequency information to adapt to different network conditions.

[0142] The separated high-frequency information is used for detail enhancement, making the naked-eye 3D picture sharp and vivid.

[0143] 3. The adaptability of naked-eye 3D technology to different devices and bandwidth environments is crucial. The light field data layering + multi-scale transformation method of this invention can optimize data transmission and ensure stable display effects under different network conditions.

[0144] Transmit base layer data to ensure basic naked-eye 3D visual effects.

[0145] Simultaneously transmitting base layer and enhancement layer data improves the clarity and depth of the 3D display.

[0146] Adjust the transmission strategy for high-frequency information in real time based on network conditions to optimize user experience.

[0147] Step S4: Adaptive Bandwidth Regulation and Data Transmission

[0148] 1. The bandwidth is predicted using an autoregressive moving average model, as shown in the following formula:

[0149] ,

[0150] in, This is the bandwidth prediction value at the current moment. This is the bandwidth measurement value at the current moment. As a smoothing factor, for The bandwidth prediction value at any given time.

[0151] Compared to traditional bandwidth monitoring methods, this solution can predict bandwidth change trends in advance and reduce the impact of sudden bandwidth drops.

[0152] 2. Based on the predicted bandwidth, this invention employs an adaptive control strategy:

[0153] like > At the same time, it transmits a combination of base layer data and enhancement layer data;

[0154] like < At the same time, basic layer data is transmitted, and data compensation is performed at the receiving end.

[0155] in, The set bandwidth threshold.

[0156] This method ensures that basic 3D display quality is maintained even when bandwidth is limited.

[0157] In glasses-free 3D data communication, bandwidth fluctuations can affect the stability of data transmission and the quality of 3D display. Traditional bandwidth monitoring methods are usually reactive and cannot cope with sudden bandwidth drops, leading to screen stuttering and frame drops. In step S4, this invention uses an autoregressive moving average model for bandwidth prediction, combined with an adaptive data control strategy, to ensure that glasses-free 3D data can be transmitted stably and efficiently under different network conditions.

[0158] 1. Traditional bandwidth monitoring methods are usually adjusted based on historical bandwidth measurements, but they have the following drawbacks:

[0159] Making adjustments after a bandwidth change can lead to short-term data loss and buffering.

[0160] Unable to prevent sudden bandwidth drops in advance, affecting the continuity of naked-eye 3D;

[0161] It is not adapted to highly dynamic network environments and cannot meet the needs of real-time data transmission.

[0162] The advantages of the autoregressive moving average model are:

[0163] Real-time prediction of bandwidth trends reduces the impact of sudden bandwidth drops;

[0164] Improve the stability of data transmission and reduce stuttering and frame drops during naked-eye 3D playback;

[0165] Suitable for dynamic network environments.

[0166] 2. In real-world network environments, bandwidth fluctuates, experiencing momentary drops and congestion. Without appropriate bandwidth management strategies, the following problems may occur:

[0167] When bandwidth is sufficient, data transmission is not fully utilized, which affects the quality of 3D display.

[0168] When bandwidth is insufficient, data transmission is interrupted, resulting in screen stuttering and distortion.

[0169] 3. Real-time transmission of naked-eye 3D data requires smooth operation under varying bandwidth conditions. The adaptive bandwidth control strategy of this invention serves the following purpose:

[0170] Ensuring high-definition 3D visual effects under high-speed network conditions;

[0171] Even when bandwidth is insufficient, it can still maintain basic naked-eye 3D display and avoid complete loss of image;

[0172] Adapt to different network environments and improve the system's versatility.

[0173] Furthermore, the bandwidth prediction and control method of the present invention can be combined with FEC forward error correction technology or adaptive video coding to further optimize the reliability of data transmission.

[0174] Step S5: Error Recovery and Light Field Reconstruction

[0175] 1. The receiver uses variational regularization to recover the lost enhancement layer data, with the following optimization objective:

[0176] ,

[0177] in, For the recovered light field data, The sampling matrix of the light field data, For the received light field data, For regularization parameters, The gradient of the light field data.

[0178] Compared to mean compensation and simple interpolation, this method can adaptively adjust the recovery weights and reduce light field reconstruction errors.

[0179] 2. This invention uses peak signal-to-noise ratio and structural similarity index to evaluate the recovery effect:

[0180] ,

[0181] ,

[0182] in, Used to evaluate the signal-to-noise ratio between the recovered data and the original data. Mean square error, The maximum pixel value. Used to assess the structural similarity between the recovered data and the original data. and The mean of the dataset. and For variance, For covariance, and It is a stabilizing factor.

[0183] Experiments show that this recovery method can significantly reduce the impact of data loss on naked-eye 3D displays and improve visual quality.

[0184] Data loss is an unavoidable problem during naked-eye 3D data transmission, especially under conditions of insufficient bandwidth and unstable transmission signals. Step S5 employs a variational regularization method for error recovery, ensuring that even with data loss, the enhancement layer data of the light field can be effectively recovered, thereby improving the quality and stability of the naked-eye 3D display. Compared to traditional simple interpolation and mean compensation methods, this method has significant advantages, accurately recovering lost data and reducing reconstruction errors.

[0185] 1. In glasses-free 3D displays, the enhancement layer contains high-frequency details of the scene. Data loss can lead to image distortion or blurring, affecting the user's immersion. The limitations of traditional methods are:

[0186] The inability to accurately recover lost data can easily introduce large reconstruction errors.

[0187] The recovery quality is not adaptive and cannot effectively handle the loss of details in different areas;

[0188] The spatial gradient and structural characteristics of the light field data cannot be fully considered, resulting in poor recovery.

[0189] This invention uses variational regularization to perform fine-grained recovery of lost enhancement layer data. The advantages of variational regularization are as follows:

[0190] The restoration process is dynamically adjusted based on the structural information of different regions to avoid oversmoothing and loss of detail.

[0191] Gradient regularization is used to fully consider the edges and details of the data during the recovery process, reducing blurring effects.

[0192] It is especially suitable for restoring richly detailed enhancement layer data and improving the clarity of naked-eye 3D images.

[0193] 2. To objectively evaluate the restoration effect, step S5 uses peak signal-to-noise ratio and structural similarity index for quality assessment:

[0194] Peak signal-to-noise ratio (PSNR) is used to evaluate the signal-to-noise ratio between the restored data and the original data, measuring the quality of image restoration. A high PNR value indicates that the error between the restored data and the original data is small, and the restoration quality is good.

[0195] The structural similarity index is used to assess the similarity between the restored data and the original data in terms of structure, texture, and brightness. This index emphasizes perceptual similarity and is a standard for evaluating the quality of naked-eye 3D images.

[0196] Peak signal-to-noise ratio (PSNR) and structural similarity index (SSI) provide a comprehensive understanding of the restoration effect, ensuring that the reconstruction quality of the light field data reaches a high standard. Especially in the case of data loss, the restored data can preserve the structure and details of the original image to the greatest extent.

[0197] 3. Error recovery is a key technology for ensuring stable naked-eye 3D display quality. By employing a variational regularization method, this invention significantly reduces the impact of lost data on naked-eye 3D displays. Specifically:

[0198] The restored enhancement layer data is realistic and accurate, improving the sharpness and detail of naked-eye 3D images;

[0199] Reduce the blurring effect caused by lost data and avoid image distortion and screen stuttering;

[0200] It enhances the user's visual experience, providing a complete 3D display effect even under conditions of limited bandwidth and unstable signal.

[0201] In summary, this invention achieves efficient and stable naked-eye 3D data communication through light field data dimensionality reduction optimization, adaptive bandwidth control, and variational regularization error recovery, providing a clear and smooth 3D visual experience even in bandwidth-constrained environments.

[0202] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A naked-eye 3D data communication method, characterized in that, Includes the following steps: Step S1: Acquire light field data through a multi-view light field acquisition device, normalize the light field data, and then use a filtering method to remove noise from the light field data. Step S2: Perform variational light field optimization on the preprocessed light field data. By constructing a light field gradient regularization model and using orthogonal basis functions to represent the light field data, the light field data is reduced to a low-dimensional light field representation. Step S3: Calculate the information distribution of low-dimensional light field data based on information entropy, divide the data into layers according to the level of information entropy, obtain the basic layer data and the enhancement layer data, and perform multi-scale transformation on the light field data to decompose the light field data into low-frequency light field components and high-frequency light field components of multiple scales. Step S4: When transmitting optical field data, the network bandwidth status is detected in real time, and the transmission of basic layer data or a combination of basic layer data and enhancement layer data is selected according to the network bandwidth status. At the receiving end, the lost enhancement layer data is estimated and compensated. Step S5: The receiving end acquires the transmitted light field data and recovers the untransmitted enhancement layer data or the data lost during transmission based on the variational regularization method, so that the finally received light field data can be used for naked-eye 3D display.

2. The naked-eye 3D data communication method according to claim 1, characterized in that, In step S1, the low-dimensional light field representation is used for subsequent data encoding; The light field data normalization process in step S1 uses the minimum-maximum normalization method to map the pixel values ​​of the light field data to a preset range. The calculation formula is as follows: ， in, The normalized light field data, This is the original light field data. and These are the minimum and maximum values ​​in the light field data. These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image. These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

3. The naked-eye 3D data communication method according to claim 2, characterized in that, The noise removal in step S1 uses a Gaussian filtering method, the mathematical expression of which is as follows: ， in, The smoothed light field data, The standard deviation is Two-dimensional Gaussian kernel, The normalized light field data, These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image. These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

4. The naked-eye 3D data communication method according to claim 1, characterized in that, The light field gradient regularization model in step S2 is optimized for dimensionality reduction using the second-order Poisson equation, and its mathematical expression is as follows: ， in, For the optimized light field data, To observe light field data, As a regularization factor, The Laplacian operator for light field data, These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image. These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

5. A naked-eye 3D data communication method according to claim 1, characterized in that, In step S3, the base layer data contains the main optical field information, and the enhancement layer data contains high-frequency detail information of the optical field. In step S4, if the network bandwidth meets the conditions for transmitting enhancement layer data, the base layer data and enhancement layer data are transmitted; if the network bandwidth cannot meet the conditions for enhancing layer data transmission, the base layer data is transmitted.

6. A naked-eye 3D data communication method according to claim 5, characterized in that, The information entropy calculation in step S3 uses the Shannon entropy formula, and the mathematical expression is as follows: ， in, The information entropy of the light field data, Let be the probability distribution of the i-th gray level in the light field data. The total number of gray levels in the light field data; The multi-scale transformation of the light field data in step S3 employs the Laplace pyramid decomposition method, and the mathematical expression is as follows: ， in, For low-frequency light field components, Let S be the high-frequency optical field component at the s-th scale, where S is the scale level. This is the original light field data. These are the spatial coordinates of the light field data, representing the two-dimensional pixel positions in the image. These are the angular coordinates of the light field data, representing the direction information of the light rays in the light field.

7. A naked-eye 3D data communication method according to claim 6, characterized in that, The real-time bandwidth status detection in step S4 uses an autoregressive moving average model for prediction, and the mathematical expression is as follows: ， in, This is the bandwidth prediction value at the current moment. This is the bandwidth measurement value at the current moment. As a smoothing factor, for The bandwidth prediction value at any given time.

8. A naked-eye 3D data communication method according to claim 7, characterized in that, The data transmission priority control in step S4 adopts an adaptive bandwidth regulation strategy, and the decision rules are as follows: like > At the same time, it transmits a combination of base layer data and enhancement layer data; like < At the same time, basic layer data is transmitted, and data compensation is performed at the receiving end; in, The set bandwidth threshold.

9. A naked-eye 3D data communication method according to claim 8, characterized in that, The data error recovery in step S5 employs a variational regularization method, the mathematical expression of which is as follows: ， in, For the recovered light field data, The sampling matrix of the light field data, For the received light field data, For regularization parameters, The gradient of the light field data.

10. A naked-eye 3D data communication method according to claim 9, characterized in that, The error recovery quality evaluation in step S5 is assessed using peak signal-to-noise ratio and structural similarity index, calculated as follows: ， ， in, Used to evaluate the signal-to-noise ratio between the recovered data and the original data. Mean square error, The maximum pixel value. Used to assess the structural similarity between the recovered data and the original data. and The mean of the dataset. and For variance, For covariance, and It is a stabilizing factor.

Images