Equipment-side intelligent media compression transmission method, device, equipment and medium

Abstract

The invention provides an equipment-side intelligent media compression transmission method and device, equipment and a medium, and the method comprises the steps: analyzing media content through a built-in neural network engine of the equipment, and extracting a content feature vector; based on the content feature vector, a compression algorithm set is dynamically selected, and the compression algorithm set is associated with the content feature vector according to a preset compression strategy mapping model; multi-stage pipeline compression is executed through a hardware encoder to generate compressed media data, and the multi-stage pipeline compression comprises a preprocessing stage, a main compression stage and a post-processing stage; a transmission process of media data is optimized based on the user behavior model and a network state prediction transmission strategy; and performing end-to-end encryption transmission on the compressed media data by using a differential privacy security protocol. According to the invention, synchronous optimization of the compression ratio, transmission energy consumption and security of the media data at the equipment end is realized, and the technical bottleneck of high-definition media transmission in a mobile environment is broken through.

Description

Technical Field

[0001] This invention relates to the field of compression technology, and in particular to a device-side intelligent media compression and transmission method, apparatus, equipment, and medium. Background Technology

[0002] With the rapid development of mobile internet and IoT technologies, media data such as images and videos generated by smartphones, wearable devices, and various embedded terminals are experiencing explosive growth. How to achieve efficient compression and low-power transmission of media data on resource-constrained devices has become a critical issue that the industry urgently needs to address. Currently, mainstream media compression technologies such as JPEG, HEVC, and VP9, ​​while possessing basic compression capabilities, typically rely on fixed rules and lack dynamic sensitivity to the characteristics of media content. For example, when dealing with different semantic content such as faces, text, and landscapes, existing methods struggle to adaptively adjust compression parameters, leading to an imbalance between compression efficiency and visual quality in complex scenarios: pursuing high compression rates easily compromises details in critical areas; conversely, emphasizing detail preservation results in excessively high data volumes.

[0003] On the other hand, existing transmission solutions generally lack the ability to intelligently predict user behavior and network environment. Traditional prefetching and differential update mechanisms are mostly based on simple heuristic rules, which cannot be dynamically optimized by combining user habits and real-time network conditions, leading to inappropriate selection of transmission timing and compression strategies, which is particularly evident in weak network environments. In addition, conventional transmission protocols lack end-to-end security protection mechanisms for media data, and data packets during the differential update process are easily stolen or tampered with, posing a risk of privacy leakage. Although some research has attempted to optimize compression or transmission through machine learning, most solutions are still limited to a single technical dimension and have failed to build a full-link collaborative processing framework covering analysis, compression, transmission, and security, thus limiting their overall performance and practicality on the device side. Summary of the Invention

[0004] The technical problem to be solved by this invention is the low compression efficiency, high transmission energy consumption and insufficient security of existing device-side media compression and transmission schemes due to the lack of content awareness, resource adaptation and end-to-end security collaboration mechanisms.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a device-side intelligent media compression and transmission method, comprising the following steps: The device analyzes media content using its built-in neural network engine and extracts content feature vectors, which include at least one of content category, texture complexity, and motion intensity. Based on the content feature vector, a set of compression algorithms is dynamically selected, wherein the set of compression algorithms is associated with the content feature vector according to a preset compression strategy mapping model; Multi-stage pipeline compression is performed by a hardware encoder to generate compressed media data. The multi-stage pipeline compression includes a preprocessing stage, a main compression stage, and a post-processing stage. Optimize the transmission process of media data based on user behavior models and network state prediction transmission strategies; The compressed media data is transmitted with end-to-end encryption using a differential privacy security protocol.

[0006] Furthermore, the step of analyzing media content and extracting content feature vectors through the device's built-in neural network engine includes: Feature vectors are extracted using a pre-trained deep learning model with embedded spatial attention modules, wherein the deep learning model is lightweighted by decomposing a standard 3×3 convolution into 1×3 and 3×1 convolutions.

[0007] Furthermore, the dynamic selection of the compression algorithm set based on the content feature vector includes: Based on the content feature vector, the complexity of the algorithm set is dynamically adjusted according to the current hardware load of the device. When the hardware load exceeds a preset threshold, a lightweight algorithm subset is automatically activated. The lightweight algorithm subset adopts a simplified network structure or reduces the number of computation layers.

[0008] Furthermore, the step of performing multi-stage pipeline compression via a hardware encoder to generate compressed media data includes: In the preprocessing stage, media data is optimized by region based on content feature vectors, and the optimization parameters of each region are dynamically adjusted. During the main compression stage, the hardware encoder is invoked to execute the selected core compression algorithm, and a hardware-level lossless compression channel is simultaneously enabled to process critical data segments. In the post-processing stage, the compressed data is adaptively quantized and adjusted to generate media data.

[0009] Furthermore, the step of performing regional optimization on media data based on content feature vectors during the preprocessing stage, and dynamically adjusting the optimization parameters for each region, includes: The face region is subjected to lossless or high-fidelity compression, the text region is subjected to edge sharpening and contrast enhancement, the background region is subjected to downsampling, and the optimization parameters of each region are dynamically adjusted according to the clarity of face features, text font size and background texture complexity in the content feature vector.

[0010] Furthermore, the step of invoking the hardware encoder to execute the selected core compression algorithm during the main compression stage, and simultaneously enabling the hardware-level lossless compression channel to process key data segments, includes: The parameters of the core compression algorithm are optimized for the characteristics of different hardware encoders. The search range and step size of motion estimation are dynamically adjusted according to the motion intensity of the video content. The core compression algorithm and hardware-level lossless compression are executed synchronously through a parallel processing architecture.

[0011] Furthermore, the optimization of media data transmission process based on user behavior models and network state prediction transmission strategies includes: Based on network condition classification, transmission quality is selected, and a probability model is used to predict transmission demand in conjunction with user behavior models. When network condition deterioration is predicted, the compression level is increased in advance.

[0012] The present invention also provides a device-side intelligent media compression and transmission apparatus, comprising: The feature vector extraction module is used to analyze media content through the device's built-in neural network engine and extract content feature vectors, wherein the content feature vectors include at least one of content category, texture complexity, and motion intensity; A compression algorithm selection module is used to dynamically select a set of compression algorithms based on the content feature vector, wherein the set of compression algorithms is associated with the content feature vector according to a preset compression strategy mapping model; A multi-stage compression module is used to perform multi-stage pipeline compression through a hardware encoder to generate compressed media data. The multi-stage pipeline compression includes a preprocessing stage, a main compression stage, and a post-processing stage. The predictive transmission optimization module is used to predict transmission strategies based on user behavior models and network conditions, and optimize the transmission process of media data. The secure transmission module is used to perform end-to-end encrypted transmission of compressed media data using a differential privacy security protocol.

[0013] The present invention also provides a computer device, the computer device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the device-side intelligent media compression and transmission method as described above.

[0014] The present invention also provides a storage medium storing a computer program, which, when executed by a processor, can implement the device-side intelligent media compression and transmission method described above.

[0015] The beneficial effects of this invention are as follows: By constructing a technology chain integrating content awareness, resource adaptation, and secure transmission, a systematic improvement in device-side media processing performance is achieved. Intelligent recognition of content features based on a neural network engine enables dynamic optimization of compression strategies according to semantic features such as faces and text, improving compression ratio by approximately 20 times compared to traditional methods while maintaining the same subjective quality. Through the collaboration of a hardware encoder pipeline and a load-aware mechanism, transmission energy consumption is reduced by approximately 40%. Combined with a differential privacy protocol and a hardware-level security module, low-latency transmission is maintained while ensuring end-to-end data security. This solution effectively overcomes the technical bottleneck of balancing quality, efficiency, and security under device-side resource constraints, providing a highly reliable solution for scenarios such as mobile healthcare and real-time communication. Attached Figure Description

[0016] The specific structure of the present invention will now be described in detail with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart of the device-side intelligent media compression and transmission method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the multi-level compression process according to an embodiment of the present invention; Figure 3 This is a block diagram of a device-side intelligent media compression and transmission apparatus according to an embodiment of the present invention; Figure 4 This is a block diagram of a multi-level compression module according to an embodiment of the present invention; Figure 5 This is a schematic block diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0020] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0021] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0022] like Figure 1 As shown, an embodiment of the present invention is: a device-side intelligent media compression and transmission method, comprising the following steps: S1. Analyze media content using the device's built-in neural network engine and extract content feature vectors, wherein the content feature vectors include at least one of content category, texture complexity, and motion intensity; Further, step S1, analyzing media content using the device's built-in neural network engine and extracting content feature vectors includes: Feature vectors are extracted using a pre-trained deep learning model with embedded spatial attention modules, wherein the deep learning model is lightweighted by decomposing a standard 3×3 convolution into 1×3 and 3×1 convolutions.

[0023] In this embodiment, a pre-trained deep learning model optimized for the device is used for feature extraction. This optimization includes two key improvements: First, the standard 3×3 convolutional layers in the model are lightweighted, decomposing them into consecutive 1×3 and 3×1 convolutions. This improvement reduces the number of parameters and computational load, successfully lowering the computational load on the device-side neural network engine by approximately 30%-40%, thus significantly improving the speed and energy efficiency of feature extraction. Second, a spatial attention module is embedded in the feature extraction stage. This module calculates the importance weight of each spatial location in the feature map, adaptively strengthening the feature representation of key regions such as faces and text while suppressing non-critical background regions. Compared to the baseline model without this module, this improvement increases the accuracy of feature extraction in key regions by 15%-20%, providing a more reliable basis for the accurate selection of subsequent compression strategies.

[0024] S2. Based on the content feature vector, dynamically select a set of compression algorithms, wherein the set of compression algorithms is associated with the content feature vector according to a preset compression strategy mapping model; Furthermore, step S2, dynamically selecting a set of compression algorithms based on the content feature vector, includes: Based on the content feature vector, the complexity of the algorithm set is dynamically adjusted according to the current hardware load of the device. When the hardware load exceeds a preset threshold, a lightweight algorithm subset is automatically activated. The lightweight algorithm subset adopts a simplified network structure or reduces the number of computation layers.

[0025] In this embodiment, the complexity of the algorithm set is dynamically adjusted based on the device's current real-time hardware load (such as neural network engine utilization and CPU utilization). When the overall hardware load exceeds a preset threshold of 70%, the system automatically activates a lightweight subset of algorithms. This subset significantly reduces computational complexity while maintaining basic compression performance by employing simplified network structures, reducing the number of computational layers, lowering image sampling rates, or narrowing the inter-frame prediction range of video. This mechanism ensures that under high load scenarios, the compression task will not cause the device system to lag or exhaust resources, maintaining the smoothness and stability of device operation. Although the compression quality may be compromised, it remains within an acceptable range.

[0026] S3. Perform multi-stage pipeline compression through a hardware encoder to generate compressed media data. The multi-stage pipeline compression includes a preprocessing stage, a main compression stage, and a post-processing stage. like Figure 2 As shown, in a specific embodiment, step S3, performing multi-stage pipeline compression through a hardware encoder to generate compressed media data, includes: S31. In the preprocessing stage, the media data is optimized by region based on the content feature vector, and the optimization parameters of each region are dynamically adjusted. In one specific embodiment, step S31, performing regional optimization on the media data based on content feature vectors during the preprocessing stage, and dynamically adjusting the optimization parameters of each region, includes: The face region is subjected to lossless or high-fidelity compression, the text region is subjected to edge sharpening and contrast enhancement, the background region is subjected to downsampling, and the optimization parameters of each region are dynamically adjusted according to the clarity of face features, text font size and background texture complexity in the content feature vector.

[0027] In this embodiment, a dynamically adaptable region optimization mechanism is implemented in the preprocessing stage. First, different content regions, such as faces, text, and backgrounds, are located based on feature analysis results. Then, targeted optimization is performed: lossless or high-fidelity compression is applied to face regions while maintaining the original sampling rate to ensure 100% detail retention; edge sharpening and contrast enhancement are applied to text regions to improve clarity; and downsampling is performed on background regions with simple textures and minimal impact. The innovation of this invention lies in the fact that the optimization parameters are not fixed but dynamically adjusted based on the detailed feature vectors extracted in step S1 (such as facial feature clarity, text font size, and background texture complexity). For example, for high-clarity faces, the redundancy of lossless compression can be appropriately reduced; for small-font text, the edge sharpening intensity is enhanced. This dynamic mechanism improves the subjective quality of preprocessed media data by 12%-18% while maintaining more reasonable data volume control.

[0028] S32. In the main compression stage, the hardware encoder is invoked to execute the selected core compression algorithm, and the hardware-level lossless compression channel is simultaneously enabled to process key data segments. In one specific embodiment, step S32, invoking the hardware encoder to execute the selected core compression algorithm during the main compression stage, and simultaneously enabling the hardware-level lossless compression channel to process key data segments, includes: The parameters of the core compression algorithm are optimized for the characteristics of different hardware encoders. The search range and step size of motion estimation are dynamically adjusted according to the motion intensity of the video content. The core compression algorithm and hardware-level lossless compression are executed synchronously through a parallel processing architecture.

[0029] In this embodiment, two collaborative optimizations were performed in the main compression stage. First, the parameters of the core compression algorithm (such as H.265) were deeply optimized based on the architectural characteristics of different hardware encoders (such as Apple Video Toolbox and Qualcomm Adreno encoder). For example, the selection priority of intra-frame prediction mode was adjusted for Apple encoders, and the search range and step size of motion estimation were dynamically adjusted for Qualcomm encoders based on motion intensity. These optimizations improved the encoding efficiency of the hardware encoders by 15%-20%. Second, the collaborative working method between the hardware-level lossless compression channel and the core compression algorithm was improved, abandoning the traditional serial processing and adopting a parallel processing architecture. By simultaneously processing the determined key data segments (such as facial feature points and text characters) in the lossless compression channel while the core algorithm compresses the media, and by using data buffers and synchronization signals to ensure consistency, the overall processing efficiency was improved by 25%-30%.

[0030] S33. In the post-processing stage, the compressed data is adaptively quantized and adjusted to generate media data.

[0031] In this embodiment, an intelligent decision-making model is used for adaptive quantization adjustment in the post-processing stage. This model constructs a multi-dimensional content complexity evaluation system. In addition to traditional gradient, texture, and motion intensity indicators, it innovatively introduces user attention indicators obtained from step S1 (such as faces and text regions having higher weights) and human visual perception threshold indicators. By weighted fusion of these multi-dimensional indicators, the model can generate a more accurate quantization parameter mapping table. For example, for highly detailed and high-attention face regions, not only are lower quantization values ​​set, but they are also fine-tuned according to the visual perception threshold to ensure optimal human visual perception; for low-attention background regions, higher quantization values ​​are used within the range allowed by the visual threshold. Compared with traditional methods based on a single indicator, this intelligent model improves the subjective visual quality of the compressed data by 10%-15% while maintaining a reasonable compression rate.

[0032] S4. Optimize the transmission process of media data based on user behavior models and network status prediction transmission strategies; In one specific embodiment, the process of optimizing media data transmission based on user behavior models and network state prediction transmission strategies includes: Based on network condition classification, transmission quality is selected, and a probability model is used to predict transmission demand in conjunction with user behavior models. When network condition deterioration is predicted, the compression level is increased in advance.

[0033] In this embodiment, the predictive transmission optimization module combines a user behavior probability model (such as a log-linear model) constructed from historical data with monitoring of real-time network conditions (5G / LTE / 3G) to predict and optimize transmission strategies. Its core is to dynamically adjust the compression level and transmission timing based on the prediction results. For example, when the system predicts, based on geographical location or network signal trends, that a device is about to enter a low-signal area such as an elevator, it proactively increases the compression level to the highest level in advance, thereby completing data preprocessing before network connectivity deteriorates. This proactive optimization strategy effectively avoids high-bandwidth transmission in congested or weak network environments, significantly reducing transmission latency and additional energy consumption due to retransmissions, resulting in an overall reduction in transmission energy consumption of up to 40%.

[0034] S5. Use a differential privacy security protocol to perform end-to-end encrypted transmission of compressed media data.

[0035] like Figure 3 As shown, the present invention also provides a device-side intelligent media compression and transmission apparatus, comprising: The feature vector extraction module 10 is used to analyze media content through the built-in neural network engine of the device and extract content feature vectors, wherein the content feature vectors include at least one of content category, texture complexity and motion intensity; Compression algorithm selection module 20 is used to dynamically select a set of compression algorithms based on the content feature vector, wherein the set of compression algorithms is associated with the content feature vector according to a preset compression strategy mapping model; The multi-stage compression module 30 is used to perform multi-stage pipeline compression through a hardware encoder to generate compressed media data. The multi-stage pipeline compression includes a preprocessing stage, a main compression stage, and a post-processing stage. The predictive transmission optimization module 40 is used to predict transmission strategies based on user behavior models and network conditions, and optimize the transmission process of media data. The secure transmission module 50 is used to perform end-to-end encrypted transmission of compressed media data using a differential privacy security protocol.

[0036] Furthermore, the feature vector extraction module 10 is specifically used for: Feature vectors are extracted using a pre-trained deep learning model with embedded spatial attention modules, wherein the deep learning model is lightweighted by decomposing a standard 3×3 convolution into 1×3 and 3×1 convolutions.

[0037] Furthermore, the compression algorithm selection module 20 is specifically used for: Based on the content feature vector, the complexity of the algorithm set is dynamically adjusted according to the current hardware load of the device. When the hardware load exceeds a preset threshold, a lightweight algorithm subset is automatically activated. The lightweight algorithm subset adopts a simplified network structure or reduces the number of computation layers.

[0038] like Figure 4 As shown, the multi-level compression module 30 further includes: Preprocessing unit 31 is used to perform regional optimization on media data based on content feature vectors during the preprocessing stage and dynamically adjust the optimization parameters of each region. The main compression unit 32 is used to call the hardware encoder to execute the selected core compression algorithm during the main compression stage, and simultaneously enable the hardware-level lossless compression channel to process key data segments. The post-processing unit 33 is used to adaptively quantize and adjust the compressed data and generate media data during the post-processing stage.

[0039] Furthermore, the preprocessing unit 31 is specifically used for: The face region is subjected to lossless or high-fidelity compression, the text region is subjected to edge sharpening and contrast enhancement, the background region is subjected to downsampling, and the optimization parameters of each region are dynamically adjusted according to the clarity of face features, text font size and background texture complexity in the content feature vector.

[0040] Furthermore, the main compression unit 32 is specifically used for: The parameters of the core compression algorithm are optimized for the characteristics of different hardware encoders. The search range and step size of motion estimation are dynamically adjusted according to the motion intensity of the video content. The core compression algorithm and hardware-level lossless compression are executed synchronously through a parallel processing architecture.

[0041] Furthermore, the predictive transport optimization module 40 is specifically used for: Based on network condition classification, transmission quality is selected, and a probability model is used to predict transmission demand in conjunction with user behavior models. When network condition deterioration is predicted, the compression level is increased in advance.

[0042] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned intelligent media compression and transmission device and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0043] The aforementioned intelligent media compression and transmission device can be implemented as a computer program, which can, for example... Figure 5 It runs on the computer device shown.

[0044] Please see Figure 5 , Figure 5 This is a schematic block diagram of a computer device provided in an embodiment of this application. The computer device 500 can be a terminal or a server. The terminal can be an electronic device with communication functions, such as a smartphone, tablet, laptop, desktop computer, personal digital assistant, or wearable device. The server can be a standalone server or a server cluster composed of multiple servers.

[0045] See Figure 5 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a system bus 501. The memory may include a non-volatile storage medium 503 and internal memory 504.

[0046] The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032 includes program instructions that, when executed, cause the processor 502 to perform a device-side intelligent media compression and transmission method.

[0047] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0048] The internal memory 504 provides an environment for the execution of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a device-side intelligent media compression and transmission method.

[0049] This network interface 505 is used for network communication with other devices. Those skilled in the art will understand that... Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device 500 to which the present application is applied. The specific computer device 500 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0050] The processor 502 is used to run a computer program 5032 stored in a memory to implement the device-side intelligent media compression and transmission method as described above.

[0051] It should be understood that in the embodiments of this application, the processor 502 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0052] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0053] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When executed by a processor, the program instructions cause the processor to perform the device-side intelligent media compression and transmission method as described above.

[0054] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0055] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0056] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0057] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0058] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0059] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A device-side intelligent media compression and transmission method, characterized in that, Includes the following steps: The device analyzes media content using its built-in neural network engine and extracts content feature vectors, which include at least one of content category, texture complexity, and motion intensity. Based on the content feature vector, a set of compression algorithms is dynamically selected, wherein the set of compression algorithms is associated with the content feature vector according to a preset compression strategy mapping model; Multi-stage pipeline compression is performed by a hardware encoder to generate compressed media data. The multi-stage pipeline compression includes a preprocessing stage, a main compression stage, and a post-processing stage. Optimize the transmission process of media data based on user behavior models and network state prediction transmission strategies; The compressed media data is transmitted with end-to-end encryption using a differential privacy security protocol.

2. The device-side intelligent media compression and transmission method as described in claim 1, characterized in that, The process of analyzing media content and extracting content feature vectors using the device's built-in neural network engine includes: Feature vectors are extracted using a pre-trained deep learning model with embedded spatial attention modules, wherein the deep learning model is lightweighted by decomposing a standard 3×3 convolution into 1×3 and 3×1 convolutions.

3. The device-side intelligent media compression and transmission method as described in claim 1, characterized in that, The dynamic selection of the compression algorithm set based on the content feature vector includes: Based on the content feature vector, the complexity of the algorithm set is dynamically adjusted according to the current hardware load of the device. When the hardware load exceeds a preset threshold, a lightweight algorithm subset is automatically activated. The lightweight algorithm subset adopts a simplified network structure or reduces the number of computation layers.

4. The device-side intelligent media compression and transmission method as described in claim 1, characterized in that, The process of performing multi-stage pipeline compression via a hardware encoder to generate compressed media data includes: In the preprocessing stage, media data is optimized by region based on content feature vectors, and the optimization parameters of each region are dynamically adjusted. During the main compression stage, the hardware encoder is invoked to execute the selected core compression algorithm, and a hardware-level lossless compression channel is simultaneously enabled to process critical data segments. In the post-processing stage, the compressed data is adaptively quantized and adjusted to generate media data.

5. The device-side intelligent media compression and transmission method as described in claim 4, characterized in that, The process of optimizing media data by region based on content feature vectors during the preprocessing stage, and dynamically adjusting the optimization parameters for each region, includes: The face region is subjected to lossless or high-fidelity compression, the text region is subjected to edge sharpening and contrast enhancement, the background region is subjected to downsampling, and the optimization parameters of each region are dynamically adjusted according to the clarity of face features, text font size and background texture complexity in the content feature vector.

6. The device-side intelligent media compression and transmission method as described in claim 4, characterized in that, The step of invoking the hardware encoder to execute the selected core compression algorithm during the main compression stage, and simultaneously enabling the hardware-level lossless compression channel to process key data segments, includes: The parameters of the core compression algorithm are optimized for the characteristics of different hardware encoders. The search range and step size of motion estimation are dynamically adjusted according to the motion intensity of the video content. The core compression algorithm and hardware-level lossless compression are executed synchronously through a parallel processing architecture.

7. The device-side intelligent media compression and transmission method as described in claim 1, characterized in that, The optimized media data transmission process based on user behavior models and network state prediction transmission strategies includes: Based on network condition classification, transmission quality is selected, and a probability model is used to predict transmission demand in conjunction with user behavior models. When network condition deterioration is predicted, the compression level is increased in advance.

8. A device-side intelligent media compression and transmission apparatus, characterized in that, include: The feature vector extraction module is used to analyze media content through the device's built-in neural network engine and extract content feature vectors, wherein the content feature vectors include at least one of content category, texture complexity, and motion intensity; A compression algorithm selection module is used to dynamically select a set of compression algorithms based on the content feature vector, wherein the set of compression algorithms is associated with the content feature vector according to a preset compression strategy mapping model; A multi-stage compression module is used to perform multi-stage pipeline compression through a hardware encoder to generate compressed media data. The multi-stage pipeline compression includes a preprocessing stage, a main compression stage, and a post-processing stage. The predictive transmission optimization module is used to predict transmission strategies based on user behavior models and network conditions, and optimize the transmission process of media data. The secure transmission module is used to perform end-to-end encrypted transmission of compressed media data using a differential privacy security protocol.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the device-side intelligent media compression and transmission method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, can implement the device-side intelligent media compression and transmission method as described in any one of claims 1 to 7.

Images