A method for dynamically configuring object attributes for digital media multi-platforms

Abstract

The application relates to the technical field of digital media processing, in particular to a kind of object attribute dynamic configuration method for digital media multi-platform.The scheme aims at the configuration problem caused by the difference of technical specifications when digital media is displayed on different platforms, generates platform feature vector containing composite features by real-time collection and analysis of each platform technical specification;Nonlinear conversion model of object basic attribute to platform specific attribute is established by using multilayer perception;Subscription-published architecture containing subscription management, message publishing and message distribution module is constructed, so that digital media object can subscribe to platform strategy change notification;When receiving the notification, trigger configuration recalculation, apply new attribute to object and update configuration information.The application can dynamically adapt to platform specification changes, improve digital media content management efficiency and display effect, has the advantages of strong universality, high configuration accuracy, etc., effectively meets the multi-platform digital media transmission demand.

Description

Technical Field

[0001] This invention relates to the field of digital media processing technology, specifically to a method for dynamically configuring object attributes across multiple digital media platforms. Background Technology

[0002] With the rapid development of digital media technology, digital media content needs to be displayed and disseminated on many different platforms, such as televisions, computers, mobile phones, and tablets. However, different platforms have significant differences in technical specifications in terms of color space, file format, compression format, file size, encryption method, frame rate, and resolution.

[0003] Traditional methods for configuring digital media object attributes are often static, requiring a one-time configuration based on platform requirements after content creation. This approach means that when platform technical specifications change, the digital media content must be recreated and reconfigured, consuming significant time and resources and resulting in inconsistent display across different platforms, failing to meet users' demands for a high-quality digital media experience. Summary of the Invention

[0004] This invention provides a method for dynamic configuration of object attributes across multiple digital media platforms. By analyzing the technical specifications of each platform in real time, a platform feature vector is generated. A nonlinear transformation model from basic object attributes to platform-specific attributes is established using a multilayer perceptron. The method also adopts a publish-subscribe model to receive platform policy change notifications and trigger configuration recalculation, thereby realizing dynamic configuration of digital media object attributes and improving the display effect and management efficiency of digital media content on different platforms.

[0005] A method for dynamically configuring object attributes across multiple digital media platforms includes:

[0006] S1. Technical Specifications Collection and Feature Vector Generation: Through web crawlers, API calls, or access to official platform documentation, technical specifications of various platforms in terms of color space, file format, compression format, file size, encryption method, frame rate, and resolution are collected in real time and this information is parsed to generate platform feature vectors.

[0007] Preferably, for color space, detailed parameters such as its standard name, color gamut range, white point coordinates, and gamma correction curve are collected; for file format, the file encoding method, extension, and supported metadata are collected; for compression format, the compression algorithm used (such as JPEG or PNG compression algorithms) and compression ratio range are understood; for file size, the maximum and minimum allowed values ​​for a single file are obtained; for encryption method, the encryption algorithm used (such as AES or RSA) and encryption key management method are specified; for frame rate, the number of frames per second for video playback is determined; and for resolution, the horizontal and vertical pixel counts of the image or video are obtained.

[0008] Preferably, the collected raw information is organized and standardized, converting data represented in different formats into a unified mathematical model or quantitative indicator. Then, a platform feature vector is constructed based on the parsed data, with each dimension of the vector corresponding to a key technical feature parameter.

[0009] Preferably, the feature vector generation process includes combining the collected features to construct new features, specifically including:

[0010] Based on feature calculation, we construct comprehensive features of frame rate and resolution, and comprehensive features of file size and compression ratio;

[0011] Among them, the combined feature of frame rate and resolution measures the information density of video content per unit time through the combination of frame rate and resolution; the combined feature of file size and compression ratio reflects the compression status of files on different platforms through the combination of file size and compression ratio.

[0012] Based on feature cross-construction, cross-features of color space, encryption method and file format, as well as cross-features of frame rate, resolution and compression format are constructed;

[0013] The cross-features of color space, encryption method, and file format comprehensively consider the three aspects of color space, encryption method, and file format to construct cross-features that reflect the commonly used file formats under a specific color space and encryption method; the cross-features of frame rate, resolution, and compression format combine frame rate, resolution, and compression format to reflect the characteristics of commonly used compression formats under different frame rates and resolutions.

[0014] Preferably, the specific method for constructing the comprehensive feature of frame rate and resolution based on feature calculation involves calculating the product of frame rate and resolution, then performing standardization processing. The result is the comprehensive feature of frame rate and resolution, defined as the information density of video content per unit time. The specific calculation formula is as follows:

[0015]

[0016] In the formula, FR is the frame rate, H is the number of horizontal pixels, V is the number of vertical pixels, and H...max V max and FR max These represent the maximum vertical pixel count, horizontal pixel count, and frame rate for the platform under consideration.

[0017] Preferably, the specific method for constructing the comprehensive feature of file size and compression ratio based on feature calculation involves calculating the product of file size and compression ratio, then performing standardization processing. The result is the comprehensive feature of file size and compression ratio, used to reflect the file compression situation under different platforms. The specific calculation formula is as follows:

[0018]

[0019] In the formula, S is the file size, CR is the file compression ratio, and S max CR is the maximum file size allowed in the platform. max It is the highest compression ratio on the platform.

[0020] Preferably, the specific method for constructing the cross-features of color space, encryption method, and file format based on feature intersection is as follows:

[0021] Let C represent the color space gamut, E represent the encryption method after tag encoding, and F = [f1, f2, ..., f] represent the file format after one-hot encoding. n Then, the combined cross-feature vector of color space, encryption method, and file format is:

[0022]

[0023] Preferably, the specific method for constructing cross-features of frame rate, resolution, and compression format based on feature cross-construction is as follows:

[0024] Let the frame rate be FR, the resolution be represented by the number of horizontal pixels H and the number of vertical pixels V, and the compression format be represented by Z after tag encoding. Then the combined cross-feature vector of frame rate, resolution, and compression format is:

[0025] S2. Establishment of Nonlinear Transformation Model: Based on the generated platform feature vectors, a multilayer perceptron model architecture is selected to establish a nonlinear transformation model from basic object attributes to platform-specific attributes.

[0026] Preferably, the multilayer perceptron model architecture consists of an input layer, a hidden layer, and an output layer. The input layer receives the basic attributes of the digital media object and the platform feature vector. The hidden layer contains multiple neurons and performs nonlinear transformation on the input information through an activation function. The output layer outputs the predicted platform-specific attributes. During training, a large amount of sample data is used, and the error gradient is calculated through the backpropagation algorithm to update the weights and biases of the network, so that the model prediction results approximate the actual platform-specific attributes.

[0027] S3. Implementation of Subscription-Publish Model: Adopt the subscription-publish model to allow digital media objects to subscribe to policy change notifications from various platforms.

[0028] Preferably, the subscription-publishing mode is implemented based on the digital media object management system. When the platform's technical specifications change, a notification containing the changed technical specification type and the specific parameter values ​​before and after the change will be sent to the system's message publishing module. The system will then push the notification to the corresponding object through the message distribution module based on the digital media object's subscription information.

[0029] S4. Configuration recalculation triggered: After receiving a platform policy change notification, the digital media object parses the change information in the notification and generates a new platform feature vector.

[0030] Preferably, after the object attributes are recalculated, the newly calculated specific attributes are applied to the digital media object, and the configuration information of the object stored in the database of the digital media object management system on the platform is updated, including the new specific attributes, the platform to which it belongs, and the update time, so as to ensure that the display and distribution of the object on the platform comply with the latest technical specifications.

[0031] Compared with the prior art, the advantages of this invention are:

[0032] Highly adaptable: This invention can respond in real time to changes in the technical specifications of various platforms, obtain change information in a timely manner through the subscription-publishing model, and trigger the recalculation of the configuration of object attributes, so that the attributes of digital media objects can dynamically adapt to the requirements of different platforms, ensuring the consistency of content and the best display effect on different platforms;

[0033] Improved efficiency: It avoids the tedious process of recreating and configuring content when platform technical specifications change, which is a traditional static configuration method. This saves a lot of time and resources. Through automated configuration recalculation, it improves the management efficiency of digital media content.

[0034] High accuracy: The nonlinear transformation model established using a multilayer perceptron can learn the complex nonlinear mapping relationship between the basic attributes of an object and platform-specific attributes, thereby more accurately converting the basic attributes of an object into specific attributes suitable for different platforms and improving the accuracy of configuration.

[0035] High versatility: This invention is applicable to various types of digital media objects and multiple different platforms, and has strong versatility. By extracting the key features of the platform and establishing a universal conversion model, it can handle the differences between different platforms in terms of color space, file format and other aspects. Detailed Implementation

[0036] In one embodiment, a method for dynamically configuring object attributes across multiple digital media platforms is provided, specifically including:

[0037] Platform technical specification collection and feature vector generation:

[0038] Collect technical specifications from various platforms:

[0039] Platform A: The color space is sRGB, the color gamut range is 0.9, the supported file formats are MP4 and JPEG, the common compression format is H.264 encoding (compression ratio range is 0.3-0.5), the maximum allowed file size is 10MB, the encryption method is AES algorithm, the video frame rate is usually 30fps, and the resolution is mainly 1920×1080.

[0040] Platform B: The color space is Adobe RGB, the color gamut range is 0.95, the supported file formats are AVI and PNG, the commonly used compression format is VP9 encoding (compression ratio range is 0.4-0.6), the maximum allowed file size is 8MB, the encryption method uses the RSA algorithm, the video frame rate is usually 60fps, and the resolution is mainly 3840×2160.

[0041] Platform C: Color space is P3, color gamut range relative value is 0.85, supported file formats include MOV and TIFF, commonly used compression format is HEVC encoding (compression ratio range is 0.2-0.4), the maximum allowed file size is 12MB, the encryption method uses SM2 algorithm, the common video frame rate is 24fps, and the resolution is mainly 1280×720.

[0042] Data preprocessing and feature extraction:

[0043] Standardizing the color gamut range of a color space can be achieved by dividing the color gamut range value of each platform by the maximum color gamut range value (assuming that the relative value of the maximum color gamut range among all possible platforms is known to be 1).

[0044] Perform one-hot encoding on the file format. For example, MP4 on platform A is encoded as [1,0] (assuming there are only two formats to compare), AVI on platform B is encoded as [1,0], and MOV on platform C is encoded as [1,0] (similarly, encoding is performed on the two formats supported by each platform).

[0045] Appropriate encoding conversions are also performed on compression formats, encryption methods, etc.

[0046] Extract numerical features such as file size, frame rate, and resolution, and then perform standardization processing.

[0047] Constructing composite features and platform feature vectors:

[0048] Frame rate and resolution combined characteristics (information density): The information density of each platform is calculated according to the formula. Taking platform A as an example, its frame rate FR = 30, resolution H = 1920, and V = 1080. Assume that the maximum frame rate FR among all platforms is... max =60, maximum horizontal pixel count H max =3840, maximum vertical pixel count V max =2160, then the information density Similarly, the information density values ​​of platform B and platform C are calculated and used as part of the composite feature;

[0049] File size and compression ratio combined characteristics (expected actual size after compression): Taking platform A as an example, assuming file size S = 5MB, compression ratio CR = 0.4, maximum file size S max =10MB (maximum file size allowed by platform A), maximum compression ratio CR max =0.5 (maximum compression ratio range for platform A), then the expected actual size after compression is... Similarly, calculate the corresponding values ​​for platform B and platform C and add them to the composite feature;

[0050] Cross-features of color space, encryption method, and file format: Taking platform A as an example, the color space gamut range is 0.9 after standardization, the encryption method AES encoding is 0, and the file format MP4 one-hot encoding is [1,0]. These are combined to form the cross-feature vector. Construct the corresponding cross-feature vectors of platform B and platform C using this method;

[0051] Cross-features of frame rate, resolution, and compression format: Taking platform A as an example, with a frame rate of 30fps, a resolution of 1920×1080, and H.264 compression format with 0 encoding, the cross-feature vector is constructed as follows: Construct the corresponding cross-feature vectors of platform B and platform C using this method;

[0052] Finally, the various basic features and composite features mentioned above are combined in a predetermined order to generate the complete platform feature vectors for each platform.

[0053] Establishment of nonlinear transformation model:

[0054] Model architecture determined: A multilayer perceptron is used to establish a nonlinear conversion model. The number of neurons in the input layer is determined based on the platform feature vector dimension (including basic and composite features) and the basic attribute dimension of the digital media object. Assuming the platform feature vector has a dimension of 20 after integration, and the object's basic attributes, such as common color values ​​and basic size, have a dimension of 5, then the input layer has 25 neurons. Two hidden layers are set: the first hidden layer has 50 neurons, using the ReLU activation function; the second hidden layer has 30 neurons, also using the ReLU function. The number of neurons in the output layer is determined based on the platform-specific attribute dimension. Assuming the platform-specific attribute dimension is 10, the output layer has 10 neurons, and the activation function is selected based on the specific output attribute characteristics, choosing an appropriate output format.

[0055] Model Training: A large amount of sample data corresponding to different digital media objects on different platforms was collected. This data covers feature vectors of different platforms, basic attributes of objects, and corresponding platform-specific attributes. Mean squared error was used as the loss function, and the Adam optimization algorithm was used for model training. An appropriate learning rate (e.g., 0.001) was set, and the number of training epochs was set to 1000. During the training process, the weight parameters of the model were continuously adjusted so that the loss function value gradually decreased and tended to converge.

[0056] Implementation of the subscription-publishing model:

[0057] A publish-subscribe architecture is established in the digital media object management system. Digital media objects indicate their intention to subscribe to policy change notifications from platforms A, B, and C by sending subscription requests to the system's subscription management module.

[0058] When Platform A updates its technical specifications, such as deciding to increase the maximum supported file size to 15MB or to change the encryption method to a new self-developed encryption algorithm (and update the corresponding encoding), Platform A encapsulates these changes into standard messages through the message publishing module and sends them to the message middleware of the digital media object management system.

[0059] Based on the previous subscription records of digital media objects, the message distribution module accurately pushes the change notification of Platform A to each digital media object that has subscribed to the platform.

[0060] Configure recalculation triggers and applications:

[0061] After receiving a policy change notification from platform A, the digital media object parses the change information in the notification and regenerates a new platform feature vector according to the method described above.

[0062] The pre-trained multilayer perceptron model is invoked, taking the basic attributes of the digital media object and the new platform feature vector as input. The model then calculates and outputs the object's specific attributes under the new technical specifications of platform A. For example, for a video object, it might recalculate video encoding parameters, frame rate adjustments, etc., to suit new file size limits, encryption methods, and other specifications.

[0063] In another embodiment, the step includes continuous monitoring and feedback optimization, specifically implemented as follows:

[0064] After the digital media object is displayed and disseminated on the platform according to the new configuration, a continuous monitoring mechanism is activated to continuously evaluate the configuration effect of the digital media object by collecting feedback information from users (such as user evaluation, playback duration, number of playback interruptions, etc.) and performance data from the system (such as network transmission rate, server load, etc.).

[0065] Based on the analysis of the collected data, if common problems such as poor image quality and playback stuttering are found, or if certain configuration attributes cause system performance bottlenecks in actual operation, the relevant information is fed back to the nonlinear conversion model training stage. Using this actual feedback data as supplementary samples, the model is further optimized and trained, and the model's parameters and structure are adjusted so that it can more accurately generate platform-specific attributes that meet the actual use scenarios and have better effects, thereby continuously improving the display quality and overall performance of digital media objects on various platforms.

[0066] These new specific attributes are applied to digital media objects, such as re-encoding video files according to new encoding parameters and adjusting image files according to new color spaces and size requirements. Simultaneously, the configuration information of the digital media object stored in the management system database on platform A is updated, recording the new specific attributes, the platform it belongs to, and the update time, ensuring that the subsequent display and distribution of the object on platform A conforms to the latest technical specifications.

[0067] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0068] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for dynamically configuring object attributes across multiple digital media platforms, characterized in that, include: S1. Technical Specifications Collection and Feature Vector Generation: Through web crawlers, API calls, or access to the platform's official documentation, technical specifications of each platform in terms of color space, file format, compression format, file size, encryption method, frame rate, and resolution are collected in real time, and this information is parsed to generate platform feature vectors. The feature vector generation process includes combining collected features to construct new features, specifically including: Based on feature calculation, we construct comprehensive features of frame rate and resolution, and comprehensive features of file size and compression ratio; The specific method for constructing the combined frame rate and resolution feature based on feature calculation involves calculating the product of frame rate and resolution, then performing standardization. The result is the combined frame rate and resolution feature, defined as the information density of video content per unit time. The specific calculation formula is as follows: ； In the formula, For frame rate, This refers to the number of horizontal pixels. This refers to the number of vertical pixels. , and These represent the maximum vertical pixel count, horizontal pixel count, and frame rate for the platform under consideration. The specific method for constructing a comprehensive feature of file size and compression ratio based on feature calculation involves calculating the product of file size and compression ratio, then performing standardization. The result is the comprehensive feature of file size and compression ratio, used to reflect the file compression situation under different platforms. The specific calculation formula is as follows: ； In the formula, For file size, This refers to the file compression ratio. This is the maximum file size allowed on the platform. It is the highest compression ratio on the platform; Among them, the combined feature of frame rate and resolution measures the information density of video content per unit time through the combination of frame rate and resolution; the combined feature of file size and compression ratio reflects the compression status of files on different platforms through the combination of file size and compression ratio. Based on feature cross-construction, cross-features of color space, encryption method and file format, as well as cross-features of frame rate, resolution and compression format are constructed; The specific method for constructing cross-features of color space, encryption method, and file format based on feature intersection is as follows: Define the color space gamut range using This indicates that the encryption method encodes the tag and then uses... This indicates that the vector after one-hot encoding of the file format is... Then the combined color space, encryption method, and file format cross-feature vector is: ； The specific method for constructing cross features of frame rate, resolution, and compression format based on feature cross-construction is as follows: Set the frame rate Resolution is measured in horizontal pixels. and vertical pixel count This indicates that the compressed format, after tag encoding, is used... The combined frame rate, resolution, and compression format cross-feature vector is then expressed as follows: ; The color space, encryption method, and file format cross-features comprehensively consider the three aspects of color space, encryption method, and file format to construct cross-features that reflect the commonly used file formats under a specific color space and encryption method; the frame rate, resolution, and compression format cross-features combine frame rate, resolution, and compression format to reflect the characteristics of commonly used compression formats under different frame rates and resolutions. S2. Establishment of nonlinear transformation model: Based on the generated platform feature vector, a multilayer perceptron model architecture is selected to establish a nonlinear transformation model from the basic attributes of the object to the specific attributes of the platform. The multilayer perceptron model architecture consists of an input layer, a hidden layer, and an output layer. The input layer receives the basic attributes of the digital media object and the platform feature vector. The hidden layer contains multiple neurons and performs nonlinear transformation on the input information through an activation function. The output layer outputs the predicted platform-specific attributes. During training, a large amount of sample data is used, and the error gradient is calculated through the backpropagation algorithm to update the weights and biases of the network, so that the model prediction results approximate the actual platform-specific attributes. S3. Implementation of Subscription-Publish Model: Adopt the subscription-publish model to allow digital media objects to subscribe to policy change notifications from various platforms; S4. Configuration recalculation triggered: After receiving a platform policy change notification, the digital media object parses the change information in the notification and generates a new platform feature vector.

2. The method for dynamic configuration of object attributes across multiple digital media platforms according to claim 1, characterized in that, After recalculating the object attributes, the newly calculated specific attributes are applied to the digital media object, and the configuration information of the object stored in the database of the digital media object management system on the platform is updated, including the new specific attributes, the platform to which it belongs, and the update time, to ensure that the display and distribution of the object on the platform comply with the latest technical specifications.

Images