Conference room high-definition video signal transmission system and method based on optical fiber KVM

Abstract

The invention discloses a conference room high-definition video signal transmission system and method based on an optical fiber KVM, and relates to the technical field of signal transmission, and the method comprises the steps of S1, signal collection and feature extraction, S2, dynamic quantization factor calculation, S3, visual lossless compression coding, S4, optical fiber channel dynamic distribution, and S5, multi-priority data packaging. According to the method, a high-detail region and a low-redundancy region in an image can be identified based on a dynamic quantization mechanism of a region entropy value, differentiated compression strength is implemented for different regions, and the method has the advantages of high resolution, high resolution, high resolution and the like. According to the method and the system, excessive compression of a high-complexity area or redundant transmission of a low-complexity area in a traditional fixed compression ratio scheme is avoided, a layered packaging technology is combined, the system can flexibly select a transmission data level according to a real-time channel condition, complete transmission of core information is preferentially guaranteed, and meanwhile enhancement and expansion layer data is loaded according to needs.

Description

Technical Field

[0001] The present invention belongs to the technical field of signal transmission, and particularly relates to a high-definition video signal transmission system and method for a conference room based on fiber optic KVM. Background Art

[0002] With the development of multimedia technology, the demand for high-definition video transmission in conference rooms is increasing day by day. Traditional KVM (Keyboard-Video-Mouse) extenders face the following technical bottlenecks when transmitting 4K / 8K ultra-high-definition signals: Most traditional solutions use general video coding such as H.264 / HEVC, and adopt a unified compression ratio for the entire frame of video, unable to recognize the differences between high-entropy regions (such as text and charts) and low-entropy regions (such as solid-color backgrounds) in the scene of mixed text and graphics, resulting in loss of details or redundant transmission. For example, when transmitting a PPT document, the edges of the text are prone to blur, while the solid-color background still occupies a large amount of bandwidth. In view of the above problems, the following solutions are proposed. Summary of the Invention

[0003] The purpose of the present invention is to provide a high-definition video signal transmission system and method for a conference room based on fiber optic KVM. By dynamically analyzing the complexity of video content, adjusting the compression strategy and bandwidth allocation in real time, the utilization rate of network resources is optimized, and the problem that the existing technology uses a unified compression ratio, resulting in loss of details or redundant transmission, is solved.

[0004] To solve the above technical problems, the present invention is realized through the following technical solutions:

[0005] The present invention is a high-definition video signal transmission system and method for a conference room based on fiber optic KVM, including:

[0006] Step S1, signal acquisition and feature extraction: Collect 4K video signals through FPGA, calculate the regional entropy value matrix of 8×8 pixel blocks, and establish the mathematical basis for video signal analysis;

[0007] Step S2, dynamic quantization factor calculation: Generate an adaptive quantization matrix based on the calculated entropy value matrix to realize dynamic adjustment of compression parameters;

[0008] Step S3, visually lossless compression coding: Adopt an improved DCT transform, combine dynamic quantization and entropy coding to balance compression efficiency and visual quality;

[0009] Step S4, dynamic allocation of fiber optic channels: According to the service weight and time-varying priority factor, monitor the channel state in real time and dynamically divide the fiber optic bandwidth resources;

[0010] Step S5, multi-priority data encapsulation: Layer-pack the video data and adopt RS coding for differential forward error correction;

[0011] Step S6, Fiber Optic Transmission Optimization: Using dual-wavelength wavelength division multiplexing technology, dynamically adjust the transmission power to ensure transmission stability;

[0012] Step S7, Receiver-side Adaptive Reconstruction: Through LSTM prediction for packet loss compensation and time-domain filtering to eliminate compression artifacts;

[0013] Step S8, Terminal Display Optimization: Apply super-resolution reconstruction algorithm to restore video details and improve terminal display quality.

[0014] Further, in step S1, signal acquisition and feature extraction specifically include the following steps:

[0015] Step S11, Signal Acquisition: Use FPGA to acquire 4K@60Hz HDMI signals;

[0016] Step S12, Feature Calculation: Calculate the regional entropy value matrix of each frame in real time. The formula is:

[0017]

[0018] In the formula, E(x,y) is the entropy value of the 8×8 pixel block with coordinates (x,y), (x,y) is the coordinate of the 8×8 pixel block, is the occurrence probability that the pixel value in the current pixel block is i (0 - 255);

[0019] In step S1, using FPGA to acquire the original HDMI signal ensures zero-compression input (retaining key information such as HDR metadata);

[0020] Quantify the local complexity of the image through the regional entropy value matrix: The higher the entropy value of the 8×8 block (such as text / texture area), the lower the data redundancy, and the compression intensity needs to be reduced;

[0021] At the same time, it provides a quantization basis for dynamic quantization, avoiding detail loss caused by uniform compression of the entire frame in traditional schemes.

[0022] Further, in step S2, the adaptive quantization matrix generated in the calculation of the dynamic quantization factor is:

[0023]

[0024] In the formula, Q adapt (x,y) is the quantization step at the position (x,y) in the adaptive quantization matrix, Q base is the basic quantization step, α is the empirical adjustment coefficient, E(x,y) is the entropy value of the 8×8 pixel block with coordinates (x,y), E avg is the average entropy value of all 8×8 blocks in the current frame;

[0025] In step S2, the change amplitude of the quantization factor is constrained by the hyperbolic tangent function (to prevent block effects caused by mutations).

[0026] When the high-entropy region (Ex,y / E_avg > 1), the quantization step size is reduced by 15% - 30% to retain edge / texture details, and the compression rate of the low-entropy region (such as a solid-color background) is automatically increased.

[0027] Furthermore, in the said step S3, the visually lossless compression encoding specifically includes the following steps:

[0028] Step S31: An improved DCT transform is adopted, and the formula is:

[0029]

[0030] In the formula, C(u,v) is the transform domain coefficient, ω(u) and ω(v) are both improved weight functions, f(x,y) is the pixel value at the coordinate (x,y) in the input image block, x and y are the spatial domain pixel coordinates, and u and v are the frequency domain coefficient coordinates;

[0031] The mathematical expression of the said improved weight function is:

[0032]

[0033] In the formula, j is the frequency domain coefficient coordinate;

[0034] Step S32: Apply the Q adapt quantization matrix to the high-frequency components, and give priority to retaining the original data in the low-entropy region during entropy encoding;

[0035] In step S3, by improving the weight coefficient of the DCT kernel function, the truncation error of the high-frequency components is reduced;

[0036] The hierarchical entropy encoding strategy can compress the data volume of the base layer to meet the transmission of low-bandwidth channels.

[0037] Furthermore, in the said step S4, the optical fiber bandwidth allocation formula in the dynamic allocation of optical fiber channels is:

[0038]

[0039] In the formula, B alloc is the bandwidth allocated to the video service at the current moment, B total is the total available bandwidth of the optical fiber, W k is the weight coefficient of the kth type of service, and S k (t) is the time-varying priority factor;

[0040] In step S4, through the time-varying priority factor S k (t) algorithm: Detect the mouse movement speed (when it is 500 pixels per second, S kInteraction features such as (+0.3) and window refresh frequency;

[0041] Service weight W k Setting example: video stream (0.6), USB data (0.3), audio (0.1);

[0042] Achieve instantaneous bandwidth expansion during burst data transmission (such as allocating an additional 200 Mbps instantly when turning the PPT page).

[0043] Furthermore, in step S5, the multi-priority data encapsulation specifically includes the following steps:

[0044] Step S51: Divide the video data into:

[0045] Base layer: Contains the DC component and motion vectors of the video blocks. When the network bandwidth is insufficient, only transmitting this layer can still maintain a viewable video;

[0046] Enhancement layer: Contains low-frequency AC components, supplementing more details on top of the base layer;

[0047] Extension layer: Contains high-frequency AC components, providing detailed information of the highest picture quality and having the greatest bandwidth requirement, transmitted when the network condition is good;

[0048] Step S52: Perform forward error correction using RS coding. Among them, the base layer is RS(255,223), the enhancement layer is RS(255,239), and the extension layer has no redundancy;

[0049] In step S5, the base layer forcibly uses forward error correction: It can correct burst errors of 16 bytes in every 255 bytes;

[0050] The extension layer uses UDP lightweight encapsulation, allowing 10 -6 levels of controllable packet loss (compensated by the LSTM at the receiving end);

[0051] The hierarchical structure enables a degraded picture to be displayed even when the transmission is interrupted (such as maintaining 1080P clarity when only the extension layer is lost).

[0052] Furthermore, in step S6, the fiber optic transmission optimization specifically includes the following steps:

[0053] Step S61: Use 1310nm / 1550nm dual-wavelength wavelength division multiplexing technology to separate the control signal and video data;

[0054] Step S62: Dynamically adjust the transmission power to ensure that the transmission loss is within a safe range. Among them, the formula for dynamically adjusting the transmission power is:

[0055]

[0056] In the formula, Ptx is the laser emission power, P min is the minimum power to ensure communication, BER target is the target bit error rate, BER current is the currently measured bit error rate in real time, K adj is the power adjustment coefficient;

[0057] In step S6, dual-wavelength wavelength division multiplexing increases the single-fiber capacity to 48 Gbps × 2, supporting 8K@60Hz redundant transmission;

[0058] The dynamic power algorithm compensates for fiber bending loss in real time (when the loss coefficient changes by ±0.02 dB per kilometer, the power is readjusted within 3 ms);

[0059] Automatically switch the coding scheme according to the change of BER (when BER > 10 -9 enable Turbo code to replace RS code).

[0060] Furthermore, in step S7, the LSTM prediction formula in the receiver adaptive reconstruction is:

[0061]

[0062] In the formula, is the predicted lost pixel block data, σ is the sigmoid activation function, W f is the LSTM network weight matrix, h t-1 is the hidden layer state of the previous time step, f missing is the known pixel value of the adjacent area of the lost block;

[0063] In step S7, the LSTM network predicts the content of the lost block by analyzing the motion vectors of the previous 5 frames;

[0064] The time-domain compensation filter eliminates the ringing effect caused by the discard of high-frequency components (applying Gaussian weighted smoothing at the 8×8 block boundary);

[0065] Dynamic metadata reorganization ensures the correct mapping of the HDR gamma curve.

[0066] Furthermore, in step S8, the super-resolution reconstruction formula in the terminal display optimization is:

[0067]

[0068] In the formula, I SR is the output super-resolution image, w i is the fusion weight of the i-th convolution kernel, D conv is the depthwise separable convolution operation, I LR is the input low-resolution image block, K iFour dedicated pre-trained convolutional kernels: K1: 3×3 sharpening kernel (enhance text edges), K2: 5×5 smoothing kernel (improve color transitions), K3: 7×7 texture kernel (restore details), K4: 1×1 correction kernel (suppress noise);

[0069] In step S8, the super-resolution convolutional kernel group {K1-K4} is optimized for text sharpening, color transition, texture enhancement, and noise suppression respectively;

[0070] Correct the sub-pixel shift caused by compression (compensate for 0.2-0.5 pixel displacement through the phase correlation algorithm);

[0071] The EDID simulation module ensures that the display device is recognized as a native 4K input, avoiding additional latency caused by scaling.

[0072] A high-definition video signal transmission system for a conference room based on fiber optic KVM, the transmission system includes a video source, a signal acquisition and feature extraction module, a dynamic quantization factor calculation module, a visually lossless compression coding module, a fiber optic channel dynamic allocation module, a multi-priority data encapsulation module, a fiber optic transmission optimization module, a receiver-side adaptive reconstruction module, a terminal display optimization module, and a terminal display;

[0073] The video source, the signal acquisition and feature extraction module, the dynamic quantization factor calculation module, the visually lossless compression coding module, the fiber optic channel dynamic allocation module, the multi-priority data encapsulation module, the fiber optic transmission optimization module, the receiver-side adaptive reconstruction module, the terminal display optimization module, and the terminal display are connected in sequence, and the output end of the fiber optic transmission optimization module is unidirectionally connected to the input end of the fiber optic channel dynamic allocation module.

[0074] Furthermore, the signal acquisition and feature extraction module is used to collect 4K@60Hz HDMI high-definition video signals using FPGA, and at the same time calculate the regional entropy value matrix for each frame of video;

[0075] The dynamic quantization factor calculation module is used to generate an adaptive quantization matrix according to the regional entropy value matrix obtained by the signal acquisition module;

[0076] The visually lossless compression coding module is used to process the video signal using an improved DCT transform, and combine the adaptive quantization matrix generated by the dynamic quantization factor calculation module to quantize the high-frequency components;

[0077] The fiber optic channel dynamic allocation module is used to monitor the status of the fiber optic channel in real time, and dynamically allocate the bandwidth resources of the fiber optic channel according to the service weight coefficient and the time-varying priority factor;

[0078] The multi-priority data encapsulation module is used to perform hierarchical processing on the compressed and encoded video data, dividing it into a base layer, an enhancement layer, and an extension layer. Forward error correction is performed on the base layer and the enhancement layer using RS coding with different parameters, and no redundancy is added to the extension layer.

[0079] The optical fiber transmission optimization module is used to separate and transmit the control signal and the video data by using the 1310nm / 1550nm dual-wavelength wavelength division multiplexing technology. At the same time, the transmission power is dynamically adjusted according to the current bit error rate.

[0080] The receiver-side adaptive reconstruction module is used to, when receiving the video data, if there is a packet loss situation, use the LSTM neural network to predict and compensate for the lost packet data.

[0081] The terminal display optimization module is used to apply the spatial domain super-resolution reconstruction algorithm to the reconstructed video data, and use the trained 4K super-resolution convolution kernel group to restore the details lost during the compression and transmission of the video.

[0082] The present invention has the following beneficial effects:

[0083] 1. By dynamically analyzing the video content complexity, the present invention adjusts the compression strategy and bandwidth allocation in real time, optimizing the utilization rate of network resources. Specifically, the dynamic quantization mechanism based on regional entropy values can identify high-detail regions and low-redundancy regions in the image, and implement different compression intensities for different regions, avoiding over-compression of high-complexity regions or redundant transmission of low-complexity regions in the traditional fixed compression ratio scheme. Combining with the hierarchical encapsulation technology, the system can flexibly select the data layer to be transmitted according to the real-time channel conditions, giving priority to ensuring the complete transmission of core information, and loading the enhancement and extension layer data as needed. This design not only reduces the dependence on fixed bandwidth but also can maintain stable video quality during network fluctuations.

[0084] 2. Through the collaborative design of hierarchical data encapsulation and forward error correction coding, the system improves the robustness of video transmission. The base layer uses high-redundancy error correction coding to ensure the complete recovery of key data under channel noise or burst interference. Specifically, the enhancement layer and the extension layer balance efficiency and reliability through different error correction strategies. The receiver combines neural network prediction and spatio-temporal filtering technologies to intelligently compensate for lost or damaged data blocks during transmission, reducing picture distortion caused by bit errors or packet losses. This design not only avoids bandwidth waste caused by excessive redundancy but also can maintain the coherence and usability of the video during physical layer damage or network congestion.

[0085] 3. Through end-to-end adaptive processing, the present invention continuously optimizes the visual experience in the compression, transmission, and reconstruction links. The dynamic quantization algorithm and improved transform coding technology at the sending end effectively reduce the excessive truncation of high-frequency details and suppress the block effect and ringing artifacts commonly seen in traditional compression. Specifically, the super-resolution reconstruction and dynamic metadata restoration technology based on deep learning at the receiving end can restore the blurred details caused by compression or transmission loss, and at the same time accurately restore the color space and dynamic range information. This design not only ensures the lossless perception of subjective vision but also can dynamically adjust the output signal according to the characteristics of the display device to avoid the delay or distortion introduced by secondary processing, ultimately achieving full-link image quality guarantee from signal acquisition to terminal presentation.

[0086] 4. Through modular design and cross-layer optimization mechanism, the present invention has broad scene compatibility and scalability. The hierarchical encapsulation structure and dynamic bandwidth allocation algorithm support seamless adaptation to video formats with different resolutions, frame rates, and color depths, and can be compatible with mainstream audio and video protocols without hardware reconstruction. Specifically, the wavelength division multiplexing and power adjustment technology of the optical fiber channel can adapt to the physical characteristics of different transmission distances and fiber types, balancing the transmission capacity and signal integrity requirements. In addition, the content-aware compression strategy can automatically optimize the processing logic according to the characteristics of common graphic-text mixing and dynamic presentations in the conference room scenario, taking into account the document clarity and video fluency. This flexibility enables it to meet the needs of diverse application scenarios from conventional video conferencing to high-fidelity medical image collaboration.

[0087] Of course, it is not necessary for any product implementing the present invention to achieve all the above-mentioned advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0089] Figure 1 It is a schematic flowchart of a method for transmitting high-definition video signals in a conference room based on fiber optic KVM of the present invention;

[0090] Figure 2 It is a framework diagram of a system for transmitting high-definition video signals in a conference room based on fiber optic KVM of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0091] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0092] Please refer to Figure 1 As shown, the present invention is a method for transmitting high-definition video signals in a conference room based on fiber optic KVM, including:

[0093] Step S1, signal acquisition and feature extraction:

[0094] Signal acquisition: The FPGA is used to acquire 4K@60Hz HDMI signals (resolution 3840×2160, YCbCr4:4:4 chrominance sampling).

[0095] Feature calculation: Calculate the regional entropy value matrix of each frame in real time. The formula is:

[0096]

[0097] In the formula, E(x,y) is the entropy value of the 8×8 pixel block with coordinates (x,y), (x,y) is the coordinate of the 8×8 pixel block, is the occurrence probability that the pixel value in the current pixel block is i (0 - 255), which is used to characterize the complexity of the video content;

[0098] Step S2, dynamic quantization factor calculation:

[0099] Generate an adaptive quantization matrix:

[0100]

[0101] In the formula, Q adapt (x,y) is the quantization step at the position (x,y) in the adaptive quantization matrix, Q base is the basic quantization step, α is the empirical adjustment coefficient, E(x,y) is the entropy value of the 8×8 pixel block with coordinates (x,y), E avg is the average entropy value of all 8×8 blocks in the current frame, and the quantization step is dynamically adjusted through the entropy value difference;

[0102] Step S3, visually lossless compression coding:

[0103] Adopt an improved DCT transform:

[0104]

[0105] Wherein, C(u, v) is the transform domain coefficient, ω(u) and ω(v) are both improved weight functions, f(x, y) is the pixel value at the coordinate (x, y) in the input image block, x and y are the pixel coordinates in the spatial domain, and u and v are the frequency domain coefficient coordinates;

[0106] Among them, the mathematical expression of the improved weight function is:

[0107]

[0108] Wherein, j is the frequency domain coefficient coordinate;

[0109] Apply Q adapt quantization matrix to the high-frequency components, and preferentially retain the original data in the low-entropy region during entropy coding;

[0110] Step S4, dynamic allocation of optical fiber channels:

[0111] Monitor the channel status in real time, and allocate bandwidth through the formula to achieve dynamic adaptation of channel resources. In the formula, B alloc is the bandwidth allocated to the video service at the current moment, B total is the total available bandwidth of the optical fiber, W k is the weight coefficient of the k-th type of service, and S k (t) is the time-varying priority factor;

[0112] Step S5, multi-priority data encapsulation:

[0113] Layer the video data: base layer (DC coefficient + motion vector), enhancement layer (low-frequency AC coefficient), extension layer (high-frequency AC coefficient);

[0114] Adopt RS coding for forward error correction (RS255,223 for the base layer, RS255,239 for the enhancement layer, no redundancy for the extension layer) to improve the transmission fault tolerance;

[0115] Step S6, optical fiber transmission optimization:

[0116] Use 1310nm / 1550nm dual-wavelength wavelength division multiplexing technology to separate the control signal and video data;

[0117] Dynamically adjust the transmission power to ensure that the transmission loss is within the safe range. The formula is:

[0118]

[0119] Wherein, P tx is the laser transmission power, P min is the minimum power to ensure communication, BER target is the target bit error rate, BER currentis the current bit error rate measured in real time, K adj is the power adjustment coefficient;

[0120] Step S7, Receiver Adaptive Reconstruction:

[0121] Packet loss compensation prediction based on LSTM neural network:

[0122]

[0123] In the formula, is the predicted lost pixel block data, σ is the sigmoid activation function, W f is the LSTM network weight matrix, h t-1 is the hidden layer state of the previous time step, f missing is the known pixel value of the adjacent area of the lost block;

[0124] Time domain compensation filtering eliminates compression artifacts and repairs transmission damage;

[0125] Step S8, Terminal Display Optimization:

[0126] Apply spatial domain super-resolution reconstruction:

[0127]

[0128] In the formula, I SR is the output super-resolution image, w i is the fusion weight of the i-th convolution kernel, D conv is the depthwise separable convolution operation, I LR is the input low-resolution image block, K i are 4 pre-trained dedicated convolution kernels;

[0129] Using the trained 4K super-resolution convolution kernel group {K i}, restore the details of compression / transmission loss and optimize the terminal display quality;

[0130] Please refer to Figure 2 As shown in, the present invention is a high-definition video signal transmission system for a conference room based on fiber optic KVM. The transmission system includes a video source, a signal acquisition and feature extraction module, a dynamic quantization factor calculation module, a visually lossless compression coding module, a fiber optic channel dynamic allocation module, a multi-priority data encapsulation module, a fiber optic transmission optimization module, a receiver adaptive reconstruction module, a terminal display optimization module, and a terminal display;

[0131] The video source, signal acquisition and feature extraction module, dynamic quantization factor calculation module, visually lossless compression coding module, optical fiber channel dynamic allocation module, multi-priority data encapsulation module, optical fiber transmission optimization module, receiver-side adaptive reconstruction module, terminal display optimization module, and terminal display are connected in sequence. The output end of the optical fiber transmission optimization module is unidirectionally connected to the input end of the optical fiber channel dynamic allocation module.

[0132] The signal acquisition and feature extraction module is used to collect 4K@60Hz HDMI high-definition video signals using an FPGA, and at the same time calculate the regional entropy value matrix for each frame of video.

[0133] The dynamic quantization factor calculation module is used to generate an adaptive quantization matrix based on the regional entropy value matrix obtained by the signal acquisition module.

[0134] The visually lossless compression coding module is used to process the video signal using an improved DCT transform, and combine with the adaptive quantization matrix generated by the dynamic quantization factor calculation module to quantize the high-frequency components.

[0135] The optical fiber channel dynamic allocation module is used to monitor the status of the optical fiber channel in real time, and dynamically allocate the bandwidth resources of the optical fiber channel according to the service weight coefficient and the time-varying priority factor.

[0136] The multi-priority data encapsulation module is used to perform hierarchical processing on the compressed and encoded video data, dividing it into a base layer, an enhancement layer, and an extension layer. Different-parameter RS coding is used for forward error correction on the base layer and the enhancement layer, and no redundancy is added to the extension layer.

[0137] The optical fiber transmission optimization module is used to use 1310nm / 1550nm dual-wavelength wavelength division multiplexing technology to separate and transmit the control signal and the video data. At the same time, the transmission power is dynamically adjusted according to the current bit error rate.

[0138] The receiver-side adaptive reconstruction module is used to, when the video data is received, if there is a packet loss situation, use an LSTM neural network to predict and compensate for the lost packet data.

[0139] The terminal display optimization module is used to apply a spatial domain super-resolution reconstruction algorithm to the reconstructed video data, and use a trained 4K super-resolution convolution kernel group to restore the details lost during compression and transmission of the video.

[0140] A specific application of this embodiment is:

[0141] Scenario: 8K video optical fiber transmission system for an intelligent conference room;

[0142] I. Hardware platform configuration

[0143] Signal transmitter: FPGA chip: Xilinx UltraScale+ VU13P (supporting 8K@60Hz HDMI 2.1 input); optical module: 100G QSFP28 (dual-wavelength WDM, 1310nm / 1550nm); storage unit: DDR4-3200 16GB (for caching video frames and entropy value matrices);

[0144] Optical fiber link: Fiber type: OM4 multimode fiber (10km transmission distance); Bit error rate benchmark: ≤1×10 -12 (BER target set value);

[0145] Signal receiver: Optical receiver: Coherent receiver (supporting dynamic power compensation); Reconstruction processor: NVIDIA Jetson AGX Orin (running LSTM compensation algorithm); Display terminal: LG 88-inch 8K OLED (HDR10+ certified, HDMI 2.1 interface);

[0146] II. Dynamic bandwidth adaptive compression implementation process

[0147] 1. Signal acquisition and feature extraction: Input signal: 8K@60Hz YCbCr 4:4:4 (original bandwidth 48Gbps);

[0148] The FPGA real-time divides the video frame into 8×8 blocks and calculates the entropy value matrix for each block:

[0149] Text area (PPT page): Entropy value E≈4.8 bits / pixel;

[0150] Background area (solid color wall): Entropy value E≈1.2 bits / pixel;

[0151] Dynamic adjustment of frequency: The processing time per frame is 0.8ms;

[0152] 2. Dynamic quantization factor allocation:

[0153] Set Q base =24, α=0.3, E avg =3.0 bits / pixel;

[0154] Quantization step size for text area:

[0155] Q adapt =24×[1+0.3×tanh(4.8 / 3.0)]≈24×1.23=29.5 (rounded up to 30);

[0156] Quantization step size for background area:

[0157] Qadapt = 24 × [1 + 0.3 × tanh(1.2 / 3.0)] ≈ 24 × 0.82 = 19.7 (rounded to 20);

[0158] Compression effect: 90% of the high-frequency components are retained in the text area, and only 15% are retained in the background area;

[0159] 3. Hierarchical encapsulation and error correction:

[0160] Data encapsulation structure (taking I-frame as an example), specifically refer to Table 1 below:

[0161]

[0162] Total compression bandwidth: 2.5 + 4.8 + 8.2 = 15.5 Gbps (compression ratio ≈ 67.7%);

[0163] 4. Fiber optic transmission optimization:

[0164] Dynamic power adjustment:

[0165] Initial transmission power: -8 dBm;

[0166] BER detected current = 5 × 10 -12 When:

[0167]

[0168] Power adjustment response time: 2.3 ms;

[0169] 5. Receiver reconstruction and display:

[0170] LSTM network prediction: Input: Pixels of the adjacent 16×16 area of the lost block; Output: Predicted block PSNR = 42.6 dB (compared with the original block);

[0171] Super-resolution reconstruction: Apply the pre-trained convolutional kernel group (K1 - K4); Output resolution: 7680×4320 (8K), VMAF score = 96.5;

[0172] III. Performance test results:

[0173] Specifically refer to Table 2 below:

[0174] Indicator This solution Traditional KVM extender End-to-end delay 7.8ms 12ms Bandwidth utilization 92％ 82％ Subjective image quality (VMAF) 96.5 88.2 Error resistance (BER) <![CDATA[1.2×10 -12 > <![CDATA[5×10 -9 >

[0175] By analyzing the content complexity of the video signal in real time, dynamically adjusting the compression ratio and fiber optic bandwidth allocation, the transmission efficiency is improved while ensuring visual losslessness.

[0176] In the description of this specification, the descriptions referring to the terms "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0177] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A method for transmitting high-definition video signals in a conference room based on fiber optic KVM, characterized in that, It includes the following steps: Step S1, Signal acquisition and feature extraction: Collect 4K video signals through FPGA, calculate the regional entropy value matrix of 8×8 pixel blocks, and establish the mathematical basis for video signal analysis; Step S2, Dynamic quantization factor calculation: Generate an adaptive quantization matrix based on the calculated entropy value matrix to achieve dynamic adjustment of compression parameters; Step S3, Visual lossless compression coding: Adopt an improved DCT transform, combine dynamic quantization and entropy coding to balance compression efficiency and visual quality; Step S4, Dynamic allocation of optical fiber channels: According to service weights and time-varying priority factors, monitor the channel status in real time and dynamically divide optical fiber bandwidth resources; Step S5, Multi-priority data encapsulation: Encapsulate video data in layers and adopt RS coding for differential forward error correction; Step S6, Optical fiber transmission optimization: Utilize the dual-wavelength wavelength division multiplexing technology to dynamically adjust the transmission power; Step S7, Adaptive reconstruction at the receiving end: Predict packet loss compensation through LSTM and eliminate compression artifacts through time-domain filtering; Step S8, Terminal display optimization: Apply the super-resolution reconstruction algorithm to restore video details.

2. The method for transmitting high-definition video signals in a conference room based on optical fiber KVM according to claim 1, wherein In step S1 mentioned above, signal acquisition and feature extraction specifically include the following steps: Step S11, Signal acquisition: Collect 4K@60Hz HDMI signals through FPGA; Step S12, Feature calculation: Calculate the regional entropy value matrix of each frame in real time, and the formula is: where E(x, y) is the entropy value of the 8×8 pixel block with coordinates (x, y), and (x, y) are the coordinates of the 8×8 pixel block. is the occurrence probability that the pixel value is i (0 - 255) in the current pixel block.

3. A high-definition video signal transmission method for a conference room based on fiber optic KVM according to claim 1, characterized in that, In step S2 mentioned above, the adaptive quantization matrix generated in dynamic quantization factor calculation is: Where Q adapt (x, y) is the quantization step at the (x, y) position in the adaptive quantization matrix, Q base is the basic quantization step, α is the empirical adjustment coefficient, E(x, y) is the entropy value of the 8×8 pixel block with coordinates (x, y), and E avg is the average entropy value of all 8×8 blocks in the current frame.

4. A method for transmitting high-definition video signals in a conference room based on fiber optic KVM according to claim 1, characterized in that, In step S3 mentioned above, visual lossless compression coding specifically includes the following steps: Step S31: Adopt an improved DCT transform, and the formula is: In the formula, C(u,v) is the transform domain coefficient, ω(u), ω(v) are both improved weight functions, f(x,y) is the pixel value at the coordinate (x,y) in the input image block, x,y are both spatial domain pixel coordinates, and u,v are both frequency domain coefficient coordinates; The mathematical expression of the improved weight function is: In the formula, j is the frequency domain coefficient coordinate; Step S32: Apply Q to the high-frequency components adapt quantization matrix, and preferentially retain the original data in the low-entropy region during entropy coding.

5. A high-definition video signal transmission method for a conference room based on fiber optic KVM according to claim 1, characterized in that, In step S4 mentioned above, the optical fiber bandwidth allocation formula in dynamic allocation of optical fiber channels is: where B alloc is the bandwidth allocated to the video service at the current moment, B total is the total available bandwidth of the optical fiber, W k is the weight coefficient of the k-th type of service, S k (t) is the time-varying priority factor.

6. A method for transmitting high-definition video signals in a conference room based on fiber optic KVM according to claim 1, characterized in that, In step S5 mentioned above, multi-priority data encapsulation specifically includes the following steps: Step S51: Divide the video data into: Base layer: Contains the DC component and motion vectors of video blocks. When the network bandwidth is insufficient, only transmitting this layer can still maintain a viewable video; Enhanced layer: Contains low-frequency AC components and supplements more details on top of the base layer; Extended layer: Contains high-frequency AC components, provides detailed information with the highest image quality, has the greatest bandwidth requirement, and is transmitted when the network condition is good; Step S52: Adopt RS coding for forward error correction. Among them, the base layer is RS(255,223), the enhanced layer is RS(255,239), and the extended layer has no redundancy.

7. A method for transmitting high-definition video signals in a conference room based on fiber optic KVM according to claim 1, characterized in that, Step S6, Optical fiber transmission optimization specifically includes the following steps: Step S61: Use the 1310nm / 1550nm dual-wavelength wavelength division multiplexing technology to separate control signals and video data; Step S62: Dynamically adjust the transmission power to ensure that the transmission loss is within the safe range. Among them, the formula for dynamically adjusting the transmission power is: Wherein, P tx is the laser emission power, P min is the minimum power to ensure communication, BER target is the target bit error rate, BER current is the currently measured bit error rate in real time, K adj is the power adjustment coefficient.

8. A method for transmitting high-definition video signals in a conference room based on fiber optic KVM according to claim 1, characterized in that, In step S7 mentioned above, the LSTM prediction formula in adaptive reconstruction at the receiving end is: In the formula, is the predicted lost pixel block data, σ is the sigmoid activation function, W f is the LSTM network weight matrix, h t-1 is the hidden layer state of the previous time step, f missing is the known pixel value of the adjacent region of the lost block.

9. A high-definition video signal transmission method for a conference room based on fiber optic KVM according to claim 1, characterized in that, In step S8, the super-resolution reconstruction formula in terminal display optimization is as follows: Wherein, I SR is the output super-resolution image, w i is the fusion weight of the i-th convolution kernel, D conv is the depthwise separable convolution operation, I LR is the input low-resolution image patch, K i are 4 pre-trained dedicated convolution kernels.

10. A high-definition video signal transmission system for a conference room based on fiber optic KVM, characterized in that, The transmission system includes a video source, a signal acquisition and feature extraction module, a dynamic quantization factor calculation module, a visually lossless compression coding module, an optical fiber channel dynamic allocation module, a multi-priority data encapsulation module, an optical fiber transmission optimization module, a receiver-side adaptive reconstruction module, a terminal display optimization module, and a terminal display; The video source, the signal acquisition and feature extraction module, the dynamic quantization factor calculation module, the visually lossless compression coding module, the optical fiber channel dynamic allocation module, the multi-priority data encapsulation module, the optical fiber transmission optimization module, the receiver-side adaptive reconstruction module, the terminal display optimization module, and the terminal display are connected in sequence, and the output end of the optical fiber transmission optimization module is unidirectionally connected to the input end of the optical fiber channel dynamic allocation module.

11. A high-definition video signal transmission system for a conference room based on fiber optic KVM according to claim 10, characterized in that, The signal acquisition and feature extraction module is used to collect 4K@60Hz HDMI high-definition video signals using FPGA, and at the same time calculate the regional entropy value matrix for each frame of video; The dynamic quantization factor calculation module is used to generate an adaptive quantization matrix according to the regional entropy value matrix obtained by the signal acquisition module; The visually lossless compression coding module is used to process the video signal using an improved DCT transform, and combine the adaptive quantization matrix generated by the dynamic quantization factor calculation module to quantize the high-frequency components; The optical fiber channel dynamic allocation module is used to monitor the state of the optical fiber channel in real time, and dynamically allocate the bandwidth resources of the optical fiber channel according to the service weight coefficient and the time-varying priority factor; The multi-priority data encapsulation module is used to perform hierarchical processing on the compressed and encoded video data, divide it into a base layer, an enhancement layer, and an extension layer, perform forward error correction on the base layer and the enhancement layer using RS coding with different parameters, and no redundancy is added to the extension layer; The optical fiber transmission optimization module is used to use the 1310nm / 1550nm dual-wavelength wavelength division multiplexing technology to separate and transmit the control signal and the video data. At the same time, the transmission power is dynamically adjusted according to the current bit error rate; The receiver-side adaptive reconstruction module is used to, when receiving video data, if there is a packet loss situation, use the LSTM neural network to predict and compensate for the lost packet data; The terminal display optimization module is used to apply the spatial domain super-resolution reconstruction algorithm to the reconstructed video data, and use the trained 4K super-resolution convolution kernel group to restore the details lost during the compression and transmission of the video.

Images