Dot matrix image data processing method and system based on cloud platform

Abstract

The invention relates to the field of image processing, and provides a dot matrix image data processing method and system based on a cloud platform, and the method comprises the steps: triggering a cloud function to notify a cloud platform to generate a new task after a user uploads a frame image and real-time weather data; the cloud platform receives the original material and carries out image processing on the frame image to generate frame image data matched with the lattice screen, and meanwhile, the cloud platform carries out semantic understanding and image rendering on the real-time weather data to generate weather rendering image data; and then the cloud platform performs semantic collaborative interaction coding analysis based on visual features on the frame image data and the weather rendering image data to obtain synthetic rendering image data, and finally the synthetic rendering image data is converted into a binary format which can be directly used by a lattice screen to obtain lattice frame data. In this way, the local resource performance bottleneck can be broken through, and meanwhile accurate adaptation and real-time interactive presentation of the image on the lattice screen can be ensured.

Description

Technical Field

[0001] The present application relates to the field of image processing, and more specifically, to a method and system for processing dot matrix image data based on a cloud platform. Background Art

[0002] With advancements in the Internet of Things (IoT) and smart display technologies, dot matrix displays are demonstrating their significant value as an effective information visualization tool in areas such as traffic management, commercial advertising, and public services. Dot matrix image data processing technology is central to the development of these low-resolution displays, particularly in smart cities, industrial automation, and retail applications. It enables the display of real-time dynamic information, such as traffic guides, advertising, and environmental monitoring data.

[0003] However, the key technical challenges involved in this process include image format conversion, resolution matching, and dynamic data integration. Traditional methods that rely on local computing resources for image parsing and rendering face significant performance bottlenecks and technical limitations. Existing dot matrix image generation solutions mostly rely on fixed templates combined with static data. This model is inefficient when processing real-time changing data sources (such as real-time weather information) because manual intervention is required to complete data collection and format adjustment, which complicates the processing flow and makes it difficult to adapt to the needs of real-time information updates and interactivity in smart city construction. More importantly, image processing systems based on local hardware configurations cannot flexibly expand computing power when dealing with high-concurrency tasks, nor can they ensure accurate image adaptation services for various types of display terminals. These problems not only increase maintenance costs, but also hinder the further development of dot matrix display technology in the field of dynamic content publishing and the improvement of its service quality.

[0004] Therefore, an optimized solution for processing bitmap image data is desired. Summary of the Invention

[0005] In response to the shortcomings of the prior art, the present application provides a cloud platform-based dot matrix image data processing method and system.

[0006] According to one aspect of the present application, a method for processing dot matrix image data based on a cloud platform is provided, comprising:

[0007] Acquire original materials uploaded by users, the original materials including frame images and real-time weather data;

[0008] After the original material is uploaded successfully, the file upload success event triggers a cloud function to notify the cloud platform of a new task;

[0009] The cloud platform obtains the original material;

[0010] The cloud platform performs image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen;

[0011] The cloud platform performs semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data;

[0012] The cloud platform synthesizes the frame image data and the weather rendering image data to obtain the original material synthesized rendering image data, including: performing semantic collaborative interaction encoding based on visual features on the frame image data and the weather rendering image data to obtain a frame image-weather rendering image visual feature interaction matrix; and obtaining the original material synthesized rendering image data based on the frame image-weather rendering image visual feature interaction matrix;

[0013] The raw material composite rendered image data is converted into a binary format that can be directly used by a dot matrix screen of a device to obtain a raw material composite rendered image dot matrix frame.

[0014] According to another aspect of the present application, a cloud platform-based lattice image data processing system is provided, comprising:

[0015] The original material acquisition module is used to acquire the original material uploaded by the user, and the original material includes frame images and real-time weather data;

[0016] The upload notification module is used to trigger a cloud function to notify the cloud platform of a new task when the original material is uploaded successfully.

[0017] A cloud platform material acquisition module is used for the cloud platform to acquire the original material;

[0018] A frame image processing module is used for the cloud platform to perform image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen;

[0019] A weather data rendering module is used on the cloud platform to perform semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data;

[0020] An image synthesis module is configured to synthesize the frame image data and the weather rendering image data on the cloud platform to obtain raw material synthesized rendering image data, wherein the image synthesis module is configured to: perform semantic collaborative interaction encoding based on visual features on the frame image data and the weather rendering image data to obtain a frame image-weather rendering image visual feature interaction matrix; and obtain the raw material synthesized rendering image data based on the frame image-weather rendering image visual feature interaction matrix;

[0021] The data format conversion module is used to convert the raw material composite rendering image data into a binary format that can be directly used by the device's dot matrix screen to obtain the raw material composite rendering image dot matrix frame.

[0022] This application has significant technical effects due to the adoption of the above technical solutions:

[0023] The cloud platform-based dot matrix image data processing method and system provided in this application first triggers a cloud function to notify the cloud platform to generate a new task after the user uploads a frame image and real-time weather data. The cloud platform then receives the original material and performs image processing on the frame image to generate frame image data adapted to the dot matrix screen. At the same time, the cloud platform performs semantic understanding and image rendering on the real-time weather data to generate weather rendering image data. Subsequently, the cloud platform performs semantic collaborative interactive coding analysis based on visual features on the frame image data and the weather rendering image data to obtain synthetic rendering image data. Finally, the synthetic rendering image data is converted into a binary format that can be directly used by the dot matrix screen to obtain dot matrix frame data. In this way, the local resource performance bottleneck can be broken through, while ensuring the accurate adaptation and real-time interactive presentation of the image on the dot matrix screen. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0025] Figure 1 Flowchart of a cloud-based lattice image data processing method according to an embodiment of the present application.

[0026] Figure 2 Flowchart of step S4 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application.

[0027] Figure 3 Flowchart of step S6 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application.

[0028] Figure 4 Flowchart of step S61 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application.

[0029] Figure 5 Flowchart of step S613 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application.

[0030] Figure 64 is a system block diagram of a cloud platform-based lattice image data processing system according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0032] Based on the technical problems raised in the above background technology, this application proposes a cloud platform-based dot matrix image data processing method. Figure 1 FIG is a flow chart of a method for processing dot matrix image data based on a cloud platform according to an embodiment of the present application. Figure 1 As shown, according to the embodiment of the present application, the dot matrix image data processing method based on the cloud platform includes: S1, obtaining the original material uploaded by the user, and the original material includes a frame image and real-time weather data; S2, after the original material is uploaded successfully, the file upload success event triggers the cloud function to notify the cloud platform that there is a new task; S3, the cloud platform obtains the original material; S4, the cloud platform performs image processing on the frame image in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen; S5, the cloud platform performs semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data; S6, the cloud platform synthesizes the frame image data and the weather rendering image data to obtain original material synthesized rendering image data; S7, converts the original material synthesized rendering image data into a binary format that can be directly used by the dot matrix screen of the device to obtain an original material synthesized rendering image dot matrix frame.

[0033] That is, traditional methods rely on local computing devices for image processing and dynamic data fusion. Their computing resources are limited by fixed hardware configurations and are difficult to elastically expand to cope with large-scale concurrent tasks. This solution uses a cloud platform to achieve dynamic scheduling and distributed processing of computing resources. It can automatically allocate computing power based on task load, thereby solving the performance bottleneck problem caused by rigid constraints on hardware resources. In addition, traditional methods rely on manual intervention in data collection and format conversion when processing multi-source dynamic data such as real-time weather, resulting in redundant data processing links and response delays. This solution uses the event triggering mechanism of cloud functions to automatically start the cloud platform's semantic understanding and image rendering process after the user uploads the original material containing real-time weather data, eliminating manual intervention and shortening the end-to-end processing time from data collection to the generation of dot matrix frames, meeting the stringent requirements of smart city scenarios for information real-time. To address the adaptation accuracy issues of different display terminals, the traditional method of superimposing static data with fixed templates cannot adapt to the pixel density and display size of diverse dot matrix screens. However, this solution uses a cloud-based image processing engine to adapt the resolution and optimize the pixel level of frame images. Combined with the dynamic synthesis of weather rendering data and original materials, the resulting binary dot matrix frame data can directly match the display characteristics of the target device, thereby improving the image output quality and compatibility in multi-terminal scenarios.

[0034] In step S1, the original material uploaded by the user is obtained, and the original material includes frame images and real-time weather data. It should be understood that the frame image, as a component of the original material, usually contains static or dynamic image content that the user expects to display on the dot matrix screen. Specifically, it can cover visual elements such as graphic design for advertising, indicator icons in traffic management, and information graphics in public services. Its expression is bitmap data with a certain resolution and color mode, carrying the main picture information that needs to be visualized through the dot matrix screen. Real-time weather data is structured data that reflects the current weather conditions. It generally includes semantic information such as weather phenomena (such as sunny, rainy, snowy, cloudy, etc.), temperature, humidity, wind speed and direction, and is used to provide real-time dynamic information related to the weather. In short, by obtaining the original material uploaded by the user, a multimodal data collection channel can be established to provide basic input for subsequent cloud platform processing processes.

[0035] In step S2, after the original material is successfully uploaded, the file upload success event triggers the cloud function to notify the cloud platform of a new task. Accordingly, the user uploading the original material is the starting point of the entire processing flow. By triggering the cloud function through the file upload success event, the system can automatically perceive the arrival of the new task without manual notification or intervention, thereby improving processing efficiency, reducing human errors, and achieving end-to-end automated processing. It is worth mentioning that the cloud function, as a lightweight computing service, can be triggered at the moment the file is successfully uploaded, and quickly pass the information of the new task to the cloud platform. In this way, the cloud platform can immediately start the corresponding processing flow and perform subsequent image processing, semantic understanding, rendering and synthesis operations on the original material to meet the system's real-time requirements, especially when processing information with strong timeliness such as real-time weather data, the processed data can be displayed on the dot matrix screen as soon as possible.

[0036] In step S3, the cloud platform obtains the original material. It should be understood that the cloud platform obtains the original material in order to achieve seamless connection between uploaded data and subsequent processing modules, ensuring the reliability of the data pipeline and the integrity of the processing flow. Specifically, the frame images and real-time weather data uploaded by the user need to be received by the cloud platform and persistently stored in the distributed file system to ensure the accessibility and consistency of the data in the subsequent processing stage. This operation includes a verification mechanism for the upload results, such as confirming that the file is not damaged during transmission by verification, and verifying whether the data format meets the predefined specifications (such as whether the frame image is a supported image type and whether the weather data contains required fields). After completing the data acquisition, the cloud platform will trigger the task queue to notify the downstream processing service and establish a task identifier to track the status of the entire processing link. In other words, the core role of this step is to convert the discrete upload action on the user side into a task entity that can be programmably processed within the cloud platform, providing a verified and persistent input source for subsequent data operations, while shielding the physical differences of the storage medium through a unified data access interface to ensure the decoupling of the processing logic and the data source.

[0037] In step S4, the cloud platform performs image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen. Specifically, Figure 2 FIG. 4 is a flow chart of step S4 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application. Figure 2 As shown, the step S4 includes: S41, scaling the frame image to the pixel size of the dot matrix screen; S42, performing color quantization processing on the scaled frame image to adapt to the color capability of the dot matrix screen to obtain the frame image data.

[0038] In step S41, the frame image is scaled to the pixel size of the dot matrix screen. It should be understood that the original resolutions of frame images from different sources may be different, and the dot matrix screen has its own specific pixel size. If scaling is not performed, the resolution of the original frame image may be much higher or lower than the resolution of the dot matrix screen, and it cannot be directly displayed correctly on the dot matrix screen. For example, a high-resolution frame image displayed on a low-resolution dot matrix screen will cause the image to be blurred and distorted, and occupy too much transmission bandwidth and storage space; a low-resolution frame image displayed on a high-resolution dot matrix screen will make the image look severely pixelated and details will be lost. Therefore, by scaling the frame image to the pixel size of the dot matrix screen, the pixel resources of the dot matrix screen can be fully utilized, avoiding problems such as stretching, deformation or blurring of the image, thereby improving the display quality of the image on the dot matrix screen and providing users with clear and accurate visual information.

[0039] In step S42, the scaled frame image is subjected to color quantization processing to adapt to the color capability of the dot matrix screen to obtain the frame image data. Accordingly, considering that the frame image may originally have rich color information, its color variety and accuracy may exceed the range that the dot matrix screen can display. Due to its hardware characteristics, the dot matrix screen can usually only display a limited number of colors. If the color quantization processing is not performed, directly displaying the frame image with rich colors on the dot matrix screen will cause color distortion and the color effect of the original image cannot be accurately restored. To this end, through color quantization processing, the colors in the scaled frame image can be mapped to the closest displayable colors according to the color capability of the dot matrix screen, thereby reducing color distortion and improving the color restoration of the image, so that the image color seen by the user on the dot matrix screen is as close as possible to the color of the original image, thereby enhancing the visual effect and information communication capability of the image.

[0040] In particular, in a specific embodiment of the present application, the implementation process of color quantization processing on a scaled frame image to adapt it to the color capabilities of a dot matrix screen to obtain the frame image data includes: first, determining the color parameters of the dot matrix screen, obtaining the maximum number of colors and color space format supported by the dot matrix screen through device specification documents or hardware parameters, which serves as the basis for color quantization. Next, parsing the scaled frame image, loading the image file using an image processing library, converting it into a digital matrix format, and extracting the color information of each pixel represented by the RGB three-channel numerical value. If the image contains an alpha channel, transparent pixels must be removed or processed first; if the image is in a non-RGB color space, it must first be converted to an RGB color space. The image color distribution is then statistically analyzed, and a global color analysis is performed on the scaled frame image. The frequency of occurrence of the RGB values ​​of all pixels is counted to generate a color histogram. To reduce computational complexity, downsampling can be performed on high-resolution or color-rich images. Furthermore, a suitable color quantization algorithm must be selected based on the number of colors on the dot matrix screen and the complexity of the image. The uniform quantization algorithm divides the RGB color space into equally spaced subintervals, each corresponding to a quantized color. By calculating the quantization interval for each channel, the pixel RGB value is mapped to the center color of the corresponding interval. The median cut algorithm, based on the color histogram, recursively segments the color space's densest regions. It first treats the entire RGB color space as a region, selects the channel with the largest range, and then cuts it by the median of the number of pixels. This process is repeated until the number of regions equals the number of colors supported by the dot matrix screen. The average color of each region is used as the representative color. The octree algorithm constructs an octree structure, with an initialized root node covering the entire RGB space. The RGB value of each pixel is then passed from the root node to the child nodes layer by layer. When the number of nodes in the tree exceeds the target number of colors, the deepest nodes with the fewest pixels are merged, and the average color of the merged nodes is used as the new node color. After quantization, a color map is generated, organizing all generated representative colors and establishing a mapping between the original pixel color and the quantized color, including the quantized color set and index mapping. Then, perform color replacement pixel by pixel, traverse each pixel of the frame image, and replace the original RGB color with the corresponding quantized color index according to the mapping table. For colors that cannot be accurately matched, use the nearest neighbor algorithm to select the closest representative color. After the color replacement is completed, verification and optimization should be carried out to compare the original image with the quantized image to check whether there is significant color distortion. If distortion is found, the quantization algorithm parameters can be adjusted or a dithering algorithm can be used to simulate more color effects. Finally, the replaced pixel index values ​​are organized and output in the format required by the dot matrix screen. If the dot matrix screen supports index color mode, the pixel index matrix is ​​output directly; if it supports fixed formats such as RGB565, the quantized RGB values ​​are converted to values ​​in the corresponding format, thereby obtaining frame image data that adapts to the color capabilities of the dot matrix screen.

[0041] In step S5, the cloud platform performs semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data. Specifically, in an embodiment of the present application, step S5 includes: semantic encoding the real-time weather data to obtain a real-time weather semantic embedding coding vector; inputting the real-time weather semantic embedding coding vector into an image rendering engine based on a generative adversarial network model to obtain the weather rendering image data. It can be understood that there are significant semantic faults and rendering quality defects in the visualization of real-time weather data. The existing technology usually superimposes weather data (such as temperature, precipitation probability) on a static image in the form of text or simple icons, lacking in-depth analysis and scenario-based expression of the semantic connotation of the data. For example, when the real-time weather data contains a "red alert for heavy rain", the traditional solution may only display a fixed-design raindrop icon, and cannot dynamically adjust the rendering effect (such as rainfall density, lightning animation intensity) according to the warning level, resulting in insufficient intuitiveness and warning effect of information transmission. Based on this, this application constructs a data-driven intelligent weather rendering mechanism through the step of "semantically encoding real-time weather data to obtain real-time weather semantic embedding coding vectors, and inputting them into an image rendering engine based on a generative adversarial network model". First, semantic coding uses natural language processing technology to parse the key semantic elements in weather data (such as weather type, intensity level, and impact range), converts them into high-dimensional vector representations, and captures implicit associations between data (such as the correlation between "heavy rain" and "reduced visibility" and "traffic control"). For example, when the input is "wind speed level 8, lasting 3 hours", the coding vector not only contains numerical features, but also embeds scenario-based impact information such as the swing amplitude of traffic signs and pedestrian safety risks. Subsequently, the image rendering engine based on the generative adversarial network (GAN) generates dynamic weather images based on the semantic coding vector. The network learns the mapping relationship between weather semantics and visual elements (such as cloud morphology and rain direction) through pre-training, and can generate high-fidelity rendering results that adapt to the display characteristics of the dot matrix screen. For example, for the semantic encoding of "heavy snow", GAN can generate dynamic images in which the density of snowflakes changes with the intensity of snowfall, and automatically adapt to the color space of the target dot matrix screen (such as rendering snowflakes as white highlights or grayscale gradient patterns).

[0042] In particular, in a specific embodiment of the present application, the implementation process of semantically encoding the real-time weather data to obtain a real-time weather semantic embedding coding vector includes: first, it is necessary to clarify the structured fields of the real-time weather data, which usually cover basic attributes such as weather phenomena, temperature, humidity, wind speed and direction, air pressure, visibility, and derived attributes such as weather warning level, perceived temperature, and precipitation probability. These data are obtained through standardized interfaces (such as meteorological APIs or local sensors) to ensure that they are input in a unified format, containing real-time values ​​or enumerated values ​​of each field, and providing a standardized data source for subsequent processing. After obtaining the data, it is necessary to clean and preprocess it to ensure input quality. For missing values, reasonable default values ​​are used to fill in the field (such as no wind direction is marked as "none") or mark unknown to avoid coding deviations; outliers are detected and corrected by statistical methods, such as eliminating temperature extremes that are beyond the reasonable range of the geographical area; unifying numerical data units, and mapping categorical data to a preset enumeration system (such as classifying "red warning for heavy rain" as a specific warning category) to ensure data format consistency. Next, feature engineering processing is performed, which is divided into categorical feature encoding and numerical feature normalization. For categorical data such as weather phenomena and warning levels, one-hot encoding is used to handle discrete features with few categories (e.g., "sunny," "rain," and "snow" are encoded as [1,0,0], [0,1,0], and [0,0,1]). Label encoding is suitable for ordered features (e.g., warning levels are mapped to ascending values ​​according to severity), and a semantic mapping table is established to handle implicit associations (e.g., the semantic association between "heavy rain" and "heavy precipitation" is strengthened through a custom offset). Numerical data (such as temperature and humidity) is scaled to a uniform range using minimum-maximum normalization to eliminate the impact of dimensional differences on subsequent encoding. For example, the temperature range is scaled to the interval [0,1]. After completing basic feature processing, feature fusion and semantic enhancement are performed. The categorically encoded discrete features and the normalized numerical features are concatenated in a fixed order to form an initial feature vector containing multidimensional information (e.g., the one-hot encoded weather phenomenon, the label-encoded warning level, and the normalized temperature and humidity features are concatenated in sequence). At the same time, domain knowledge is injected to strengthen semantic expression. For example, when "weather phenomenon = heavy rain" and "warning level = red", a preset weight is added to the corresponding feature position to highlight the high-risk semantics. Subsequently, a semantic encoding model is constructed and trained to map the fused feature vector into a low-dimensional semantic embedding vector. The model is selected based on the data scale and semantic complexity: a fully connected neural network is used for small-scale data or simple semantic scenarios, and nonlinear mapping is learned through a multi-layer perceptron; a Transformer encoder is used for complex scenarios, with the help of a self-attention mechanism to capture the dependencies between multiple features (such as the relationship between temperature and perceived temperature, and precipitation probability and humidity).Training strategies are selected based on actual needs, including supervised learning (optimizing the loss function using semantic labels annotated with historical data), unsupervised learning (compressing features through autoencoders), or transfer learning (fine-tuning based on pre-trained meteorological models) to ensure that the model accurately captures the semantic connotations of weather data in the target scenario. Preprocessed real-time weather data is input into the model in the training data format to generate a fixed-dimensional semantic embedding vector. The input must strictly follow the feature order and normalization rules. The model service is called through a real-time API, and the final output is a real-time weather semantic embedding encoding vector containing semantic information such as weather phenomena, intensity, and impact range, which provides the core input for semantic-driven subsequent image rendering.

[0043] In step S6, the cloud platform synthesizes the frame image data and the weather rendering image data to obtain raw material synthesized rendering image data. It should be understood that the frame image data and the weather rendering image data come from different processing flows. The frame image data is image data suitable for display on a dot matrix screen obtained after processing the original frame image, while the weather rendering image data is image data with a visualization effect generated based on real-time weather data. In order to fully display the content containing images and weather information on the dot matrix screen, it is necessary to synthesize these two types of data. In addition, the frame image may contain various scene information, such as people, scenery, etc., while the weather rendering image data provides current weather conditions information. Synthesizing the two can make the displayed content richer and more comprehensive, allowing users to intuitively understand the weather conditions at the time while viewing the image, thereby achieving information complementarity. However, traditional methods usually use static template overlay or mechanical layer overlay to fuse multi-source data, resulting in serious lack of synergy between dynamic information and static images. For example, in smart transportation scenarios, when real-time weather warning data (such as heavy rain or fog) needs to be synthesized with traffic route maps, traditional solutions often overlay weather icons at preset positions on the frame image. This makes it impossible to dynamically adjust the rendering position and transparency based on road network density and visual focus areas, which can easily cause key traffic signs to be obscured or cause information overload. Furthermore, due to a lack of in-depth understanding of image semantics, the visual expression of weather data lacks a logical connection to the content of the frame image. For example, in commercial advertising scenarios, temperature data may be rigidly overlaid on the facial area of ​​​​the advertising character, undermining the overall aesthetics of the image and the efficiency of information communication.

[0044] Based on this, in the cloud platform, the frame image data and the weather rendering image data are synthesized to obtain the original material synthetic rendering image data. The technical concept of this application is to build a data-driven intelligent synthesis framework by introducing visual feature semantic collaborative coding and diffusion model synthesis engine. First, visual features are extracted from the frame image and the weather rendering image respectively to capture the deep semantic information such as the spatial layout and color distribution of the two; then, dynamic associations between features are established through visual feature interaction (such as semantic matching of the dense raindrop features of the rainstorm icon with the low-lying areas of the road in the frame image), and an interaction matrix reflecting the intrinsic association of the data is generated; finally, the image synthesis engine based on the diffusion model, guided by the interaction matrix, achieves the natural fusion of weather data and frame images while retaining the key information of the original image. This technical path not only solves the problems of information occlusion and visual conflict in traditional solutions, but also realizes the automation of cross-terminal adaptation through cloud computing power, providing a highly robust synthesis solution for the release of dynamic dot matrix information.

[0045] Specifically, Figure 3 FIG. 1 is a flow chart of step S6 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application. Figure 3 As shown, the step S6 includes: S61, performing semantic collaborative interaction encoding based on visual features on the frame image data and the weather rendering image data to obtain a frame image-weather rendering image visual feature interaction matrix; S62, obtaining the original material synthetic rendering image data based on the frame image-weather rendering image visual feature interaction matrix.

[0046] In step S61, the frame image data and the weather rendering image data are subjected to semantic collaborative interaction encoding based on visual features to obtain a frame image-weather rendering image visual feature interaction matrix. Specifically, Figure 4 FIG is a flow chart of step S61 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application. Figure 4 As shown, the step S61 includes: S611, performing visual feature extraction on the frame image data to obtain a frame image visual feature encoding feature vector; S612, performing visual feature extraction on the weather rendering image data to obtain a weather rendering image visual feature encoding vector; S613, performing performance-intrinsic visual feature semantic collaborative encoding on the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector to obtain the frame image-weather rendering image visual feature interaction matrix.

[0047] In step S611, visual feature extraction is performed on the frame image data to obtain a frame image visual feature encoding feature vector. Specifically, in an embodiment of the present application, step S611 includes: performing visual feature extraction based on a feature pyramid network on the frame image data to obtain the frame image visual feature encoding feature vector. Accordingly, considering that the frame image content contains rich detail information and multi-level understanding, traditional feature extraction methods (such as single-scale convolutional neural networks) can only capture local or global features of the image and cannot simultaneously represent semantic information at different levels (such as background contours, mid-ground object details, foreground key elements). For example, when the frame image contains distant buildings and nearby road signs, traditional methods may only extract overall tonal features and ignore the shape and texture information of the nearby road signs, resulting in the synthesized weather icon destroying the recognizability of the road signs. To this end, the present application performs visual feature extraction on the frame image data to obtain a frame image visual feature encoding feature vector. In particular, in a specific example of the present application, visual feature extraction based on a feature pyramid network is performed on the frame image data to obtain the frame image visual feature encoding feature vector. That is, the feature pyramid network (FPN) simultaneously extracts high-dimensional semantic features (such as scene categories, object positions) and low-dimensional detail features (such as edge direction, texture gradient) in frame image data through top-down and lateral connection structures. For example, when processing traffic guidance frame images, high-level features can identify the road grid distribution, mid-level features capture the direction of lane lines, and bottom-level features retain the jagged details of arrow signs. This multi-scale feature encoding mechanism enables the subsequent semantic collaborative encoding module to accurately judge the visual correlation between weather rendering data (such as fog effects) and road elements-rendering the fog concentration gradient along the direction of the lane lines rather than simply covering the entire picture. Compared with traditional single-scale feature extraction, the encoded feature vector generated by FPN contains both the semantic intent of "priority warning for icy roads" and the pixel-level information of "the edge sharpness of curve signs needs to be maintained."

[0048] In step S612, visual feature extraction is performed on the weather rendering image data to obtain a weather rendering image visual feature coding vector. Accordingly, it is considered that the weather rendering image data has its own unique visual features, such as color, texture, shape, etc. under different weather conditions. For example, an image on a sunny day may be dominated by bright tones and fewer textures, while an image on a rainy day may contain dense rain line textures and darker tones. Therefore, in order to accurately capture these weather-related features, the weather information is converted into a feature vector form that can be understood and processed by a computer for subsequent analysis and processing. The present application performs visual feature extraction on the weather rendering image data to obtain a weather rendering image visual feature coding vector. In particular, in a specific example of the present application, extraction can also be performed through a feature pyramid network to obtain a weather rendering image visual feature coding vector containing rich visual information.

[0049] In step S613, the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector are subjected to performance-inherent visual feature semantic collaborative encoding to obtain the frame image-weather rendering image visual feature interaction matrix. Specifically, Figure 5 FIG is a flow chart of step S613 in the cloud platform-based dot matrix image data processing method according to an embodiment of the present application. Figure 5 As shown, the step S613 includes: S613-1, performing kernel encoding based on feature principal components on the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector to obtain a set of frame-weather rendering image visual feature principal component kernel encoding vectors; S613-2, calculating the performance operator between any two frame-weather rendering image visual feature principal component kernel encoding vectors in the set of frame-weather rendering image visual feature principal component kernel encoding vectors to obtain a frame-weather rendering image visual feature performance operator topology matrix; S613-3, performing graph convolution encoding on the set of frame-weather rendering image visual feature principal component kernel encoding vectors and the frame-weather rendering image visual feature performance operator topology matrix to obtain the frame image-weather rendering image visual feature interaction matrix.

[0050] It's understandable that the feature vectors of frame images and weather renderings each contain rich but independent information. Therefore, to provide precise guidance for subsequent image synthesis, the two need to be fused. However, traditional fusion of frame images and weather rendering data often relies on simple linear correlation analysis or manually pre-set overlay rules, making it difficult to effectively capture the deep semantic connections between multi-source data. For example, when heavy rain rendering data needs to be synthesized with a traffic route map, traditional methods simply overlay layers through coordinate alignment, failing to consider the physical correlation between rainfall distribution and road topology (such as areas prone to waterlogging in bridges and tunnels). This results in the inability to accurately locate warning information in high-risk areas. This shallow connection stems from multicollinearity and redundant noise interference between feature encoding vectors. The original feature vectors are mixed with a large amount of irrelevant details (such as the frame image background texture and decorative elements of the weather icon), causing key semantic relationships (such as "heavy rain intensity" and "road visibility") to be submerged in high-dimensional noise. Based on this, the present application performs performance-intrinsic visual feature semantic collaborative encoding on the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector to obtain the frame image-weather rendering image visual feature interaction matrix.

[0051] Specifically, principal component analysis (PCA) is first performed on the frame image and weather rendering feature vectors to remove redundant noise and decouple core semantic components (such as road skeleton features in traffic route maps and precipitation intensity gradients in rainstorm rendering data). For example, in commercial advertising scenarios, PCA can extract brand logo outline features from frame images and separate numerical trend features from temperature rendering data. Subsequently, PCA is used to map the decoupled principal component vectors into a high-dimensional nonlinear space through PCK correlation encoding to learn their underlying association patterns (such as the high temperature value and the need for enhanced color contrast in cold drink advertisements). In this process, adaptive weight adjustment is used to capture complex dependencies between features (such as the nonlinear mapping between typhoon path vectors and the position offsets of warning signs on coastal roads). Finally, a correlation topology matrix is ​​constructed based on performance operators to quantify the synergistic effects between feature components (such as the visual warning priority of fog concentration and traffic flow monitoring data). A graph convolutional network is then used to perform global information propagation and aggregation, generating an interaction matrix reflecting cross-modal semantic associations.

[0052] Specifically, in the embodiment of the present application, step S613-1 includes: performing principal component analysis on the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector to obtain a set of frame image visual feature principal component encoding feature vectors and a set of weather rendering image visual feature principal component encoding vectors. This process can be expressed as follows:

[0053]

[0054] Among them, V1 is the eigenvector of the visual feature encoding of the frame image, PCA(·) is the principal component analysis of the feature, C1 is the covariance matrix of the visual feature samples of the frame image calculated by V1, X is the set of the principal component encoding eigenvectors of the visual feature of the frame image, x1, x2, x i and x m are the first, second, i-th and m-th principal component encoding feature vectors of the frame image visual feature principal component encoding feature vectors, Λ1 is the frame image visual feature diagonal matrix, λ 11 ,λ 1m x1 and x m The corresponding eigenvalues, T is the transposition operation, V2 is the weather rendering image visual feature encoding vector, C2 is the weather rendering image visual feature sample covariance matrix calculated by V2, Y is the set of weather rendering image visual feature principal component encoding vectors, y1, y2, y i and y m are the first, second, i-th and m-th principal component coding vectors of the visual features of weather rendering images in the set of principal component coding vectors of the visual features of weather rendering images, Λ2 is the diagonal matrix of the visual features of weather rendering images, λ21 ,λ 2m y1 and y m The corresponding eigenvalues;

[0055] Each corresponding set of the principal component encoding feature vectors of the frame image visual features and the principal component encoding vectors of the weather rendering image visual features in the set of the principal component encoding feature vectors of the frame image visual features and the set of the principal component encoding vectors of the weather rendering image visual features is subjected to principal component kernel encoding to obtain the set of principal component kernel encoding vectors of the frame-weather rendering image visual features. This process can be expressed as follows:

[0056]

[0057] Among them, ‖·‖ is the Euclidean norm of the calculated vector, α and β are x i and y i The corresponding weighting coefficient, v i It is the i-th frame-weather rendering image visual feature principal component kernel encoding vector in the set of frame-weather rendering image visual feature principal component kernel encoding vectors.

[0058] It should be understood that the original frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector are mixed with high-dimensional noise and redundant information, such as irrelevant background textures in the frame image and decorative elements in the weather icon. These interference items will make it difficult to effectively extract core semantic associations (such as precipitation intensity gradient and road topology). Through the orthogonal transformation operation of feature principal component analysis, the original high-dimensional features can be projected into a low-dimensional principal component space to achieve feature decoupling and noise stripping. For example, in the scene of heavy rain and traffic route map, this step can separate the road skeleton features and the principal components of precipitation intensity gradient, and eliminate the collinear interference between the precipitation range and the frame image background color in the weather rendering data. In other words, this processing process not only reduces the data dimension, but also ensures that the subsequent association analysis focuses on the core features with physical meaning by retaining the components with the largest explanatory variance, providing a de-redundant input basis for cross-modal semantic alignment.

[0059] Accordingly, traditional coordinate alignment or linear superposition can only capture shallow spatial relationships and cannot model complex physical associations. Through the principal component kernel association coding operation, the principal component coding feature vectors of the frame image visual features and the principal component coding vectors of the weather rendering image visual features (such as the brand logo outline and the temperature change trend) can be mapped to a high-dimensional nonlinear space, and the deep network can be used to learn implicit interaction patterns (such as the need to enhance the contrast between high temperature values ​​and the color of cold drink advertisements). In other words, this processing step breaks through the linear inseparability limitation through nonlinear transformation, and can make the potential associations between low-dimensional principal components separable in high-dimensional space. The intermediate representation generated by the kernel association coding (i.e., the set of principal component kernel coding vectors of the frame-weather rendering image visual features) can provide a vector space foundation with both distinguishability and semantic interpretability for the subsequent graph structure construction.

[0060] Specifically, in the embodiment of the present application, step S613-2 includes: calculating a performance operator between any two frame-weather rendered image visual feature principal component kernel encoding vectors in the set of the frame-weather rendered image visual feature principal component kernel encoding vectors to obtain a frame-weather rendered image visual feature performance operator topological moment. This process can be expressed as follows:

[0061]

[0062] Among them, v j is the jth frame-weather rendering image visual feature principal component kernel encoding vector in the set of frame-weather rendering image visual feature principal component kernel encoding vectors, v i,k and v j,k v i and v j The kth eigenvalue in v, n is i and v j The number of eigenvalues, s(v1,v1), s(v i ,v j ), s(v m ,v1),s(v1,v m ) and s(v m ,v m ) are the performance operators at each position in the topological matrix of the visual feature performance operator of the frame-weather rendering image, M A It is the frame-weather rendered image visual feature performance operator topology matrix.

[0063] It should be understandable that the synergistic effect of cross-modal features needs to be expressed quantitatively, but traditional manually preset rules (such as layer transparency overlay) cannot dynamically adapt to complex scenarios. The performance operator can quantify the strength of the association between features through a custom measurement function. For example, the physical association of "rainstorm intensity-bridge location" is mapped to a high-weight value in the matrix, while the irrelevant combination of "background texture-precipitation icon" is given a low weight. In other words, the generated frame-weather rendering image visual feature performance operator topology matrix can encode the semantic association strength as the adjacency relationship of the graph structure. For example, in traffic scenarios, the performance operator values ​​of bridge nodes and heavy precipitation areas are significantly higher than those of ordinary road sections, thereby forming local high-density connections in the matrix. This dynamic quantization mechanism can provide structured prior knowledge for graph convolution.

[0064] Specifically, in an embodiment of the present application, step S613-3 includes: performing graph convolution encoding on the set of principal component kernel encoding vectors of the frame-weather rendered image visual features and the frame-weather rendered image visual feature performance operator topology matrix to obtain the frame image-weather rendered image visual feature interaction matrix. This process can be expressed as follows:

[0065]

[0066] Among them, GCN is graph convolutional coding, m is the number of vectors in the set of principal component kernel encoding vectors of the visual features of the frame-weather rendered image, and M c is the frame image-weather rendering image visual feature interaction matrix.

[0067] It should be understandable that a single node feature (such as the visibility parameter of a certain road section) cannot independently reflect the system-level risk, and multi-hop neighbor information needs to be aggregated through multi-layer message passing of graph convolutional coding. For example, in a foggy weather scenario, the low visibility feature of a node can trigger the adaptive adjustment of the transparency of the adjacent road section nodes after being propagated through the graph convolution layer, and at the same time generate a graded warning strategy based on the global traffic flow characteristics. That is, the graph convolutional coding operation can gradually integrate local feature interactions (such as rainfall distribution and drainage manhole cover location) and global semantic constraints (such as urban traffic network topology) by iteratively updating the node representation. The final output frame image-weather rendering image visual feature interaction matrix not only contains fine-grained feature associations (such as the probability of waterlogging at a specific intersection), but also implies cross-regional collaborative strategies (such as the coordinated adjustment of the diversion arrow direction in adjacent areas), which can provide the image synthesis engine with a decision basis covering the micro to macro levels.

[0068] Preferably, in another example of the present application, the step S613-3 includes: based on the frame-weather rendering image visual feature performance operator topology matrix, performing cross-section compactification tuning based on dynamic gauge field coupling on each frame-weather rendering image visual feature principal component kernel coding vector in the set of frame-weather rendering image visual feature principal component kernel coding vectors to obtain a set of frame-weather rendering image visual feature principal component kernel optimization coding vectors; performing graph convolution coding on the set of frame-weather rendering image visual feature principal component kernel optimization coding vectors and the frame-weather rendering image visual feature performance operator topology matrix to obtain the frame image-weather rendering image visual feature interaction matrix.

[0069] Specifically, in the calculation of the graph convolution, the spatial subordination follows the distribution paradigm as follows: each frame-weather rendering image visual feature principal component kernel encoding vector v i It will serve as the basis for generating the spatial transition canonical potential field. Since the principal component kernel encoding vector v of the visual features of the frame-weather rendered image i and frame-weather rendering image visual feature performance operator topology matrix M A There is a dimensional discontinuity between i ∈R 1×n With M A ∈R n×m dimensional mismatch), it is necessary to construct the frame-weather rendering image visual feature cross-section mapping matrix M Bi ∈R n×m To establish a dimensional transition channel.

[0070] In order to suppress the distribution disorder effect caused by the random potential field, the principal component kernel encoding vector v of the visual features of the frame-weather rendering image is firstly transformed into i With v i The corresponding initialized frame-weather rendering image visual feature cross-section mapping matrix M Bi Perform matrix contraction operation to generate v i Corresponding frame-weather rendering image visual feature cross-section compact feature vector v Bi ,

[0071] Furthermore, by splicing the visual features of each frame-weather rendering image principal component kernel encoding vector v i Corresponding frame-weather rendering image visual feature cross-section compact feature vector v Bi , construct the frame-weather rendering image visual feature cross-section compactification matrix M d =(v B1 T ,v B2 T ,…,v BiT ,…,v Bm T ).

[0072] Based on this, the frame-weather rendering image visual feature cross-section compactification matrix M can be calculated d and frame-weather rendering image visual feature performance operator topology matrix M A Gaussian correlation between:

[0073]

[0074] Among them, σ 2 is the frame-weather rendering image visual feature cross-section compactification matrix M d The frame-weather rendering image visual feature performance operator topology matrix M A The variance of the set of all matrix values ​​​​of .

[0075] The optimization process iteratively updates v i The corresponding initialized frame-weather rendering image visual feature cross-section mapping matrix M Bi , and finally the distribution topology disorder phenomenon is eliminated, so that the principal component kernel encoding vector of the visual features of the frame-weather rendering image satisfies:

[0076]

[0077] This method effectively overcomes the interference of random potential field mapping on the topological space and significantly improves the geometric consistency of graph neural networks in non-Euclidean data modeling.

[0078] Subsequently, the set of principal component kernel optimized encoding vectors of the frame-weather rendered image visual features and the frame-weather rendered image visual feature performance operator topology matrix are graph convolutionally encoded to obtain the frame-weather rendered image visual feature interaction matrix. It should be understood that the optimized set of principal component kernel encoding vectors of the frame-weather rendered image visual features ensures dimensional continuity and distribution order, ensuring strict geometric consistency between the interactions between node features and topological structures during the graph convolution process. After eliminating dimensionality mismatch and random potential field interference, the feature cross-section compaction vector can accurately represent the core components of cross-modal semantic associations (such as the physical coupling relationship between precipitation gradient and road visibility). This optimization mechanism provides highly structured input for graph convolution, enabling the multi-layer message passing of graph convolution to aggregate high-order dependencies of cross-modal features with less noise interference. The resulting interaction matrix not only retains the fine-grained semantics of the original data but also implicitly models indirect causal chains in complex scenarios, thereby improving the robustness and interpretability of cross-domain feature fusion.

[0079] In step S62, based on the frame image-weather rendered image visual feature interaction matrix, the raw material composite rendered image data is obtained. Specifically, in an embodiment of the present application, step S62 includes inputting the frame image-weather rendered image visual feature interaction matrix into a diffusion model-based image synthesis engine to obtain the raw material composite rendered image data. It should be understood that synthesizing a frame image and a weather rendered image is a complex task that requires comprehensive consideration of multiple factors, such as their semantic information, spatial layout, and color fusion. Diffusion models possess powerful capabilities in image synthesis, learning the probability distribution of images to generate high-quality, diverse, and semantically logical images. Using the visual feature interaction matrix as input, the diffusion model can be guided to synthesize images based on the semantic associations between the frame image and the weather rendered image, fully leveraging the diffusion model's ability to model image details and global structure to achieve natural and accurate image synthesis. Specifically, the diffusion model-based image synthesis engine processes as follows: First, the frame image-weather rendered image visual feature interaction matrix is ​​input into the diffusion model as a prior semantic condition. This interaction matrix carries the fine-grained semantic associations and global interaction patterns of cross-modal features. The processing of the diffusion model is divided into two stages: forward diffusion and reverse denoising. The reverse denoising stage is the core process of synthesizing images: at the initial moment, the model receives random noise that conforms to the Gaussian distribution as input, and encodes the interaction matrix into a conditional vector and integrates it into each layer of the network structure to guide the mapping of noise to image. In each denoising step, the model predicts the noise distribution at the current moment through a neural network (such as the U-Net architecture), combines the semantic association information between the frame image and the weather rendering image contained in the interaction matrix, adjusts the mean and variance of the noise distribution, gradually removes the noise and restores the image structure. Specifically, the feature association weights in the interaction matrix will affect the model's attention allocation to features in different regions. For example, when synthesizing traffic route maps and rainstorm rendering images, the association weights between the "precipitation intensity gradient" and the "road prone to waterlogging areas" in the matrix will guide the model to enhance the detail generation of warning textures at corresponding locations. Through multi-step iterative denoising operations, the model ultimately outputs synthetic image data containing the main content of the frame image and weather rendering semantic information. This data adapts to the resolution and color capabilities of the dot matrix screen at the pixel-level accuracy, and achieves deep fusion of multi-source data at the semantic level. It can ensure that the synthetic image retains the visual subject of the original frame image and accurately embeds the visual expression of real-time weather data, providing a high-quality image foundation for subsequent conversion into a binary format usable by the dot matrix screen.

[0080] In summary, step S6 is clearly explained. It first extracts visual features from both the frame image and the weather rendering image to capture deep semantic information such as their spatial layout and color distribution. Then, through visual feature interaction, dynamic associations between features are established to generate an interaction matrix reflecting the inherent associations in the data. Finally, the diffusion model-based image synthesis engine, guided by the interaction matrix, achieves a natural fusion of weather data and the frame image while preserving the key information of the original image. This effectively resolves the issues of information occlusion and visual conflict in traditional solutions, and enables automated cross-terminal adaptation through cloud computing power, providing a highly robust synthesis solution for dynamic dot matrix information release.

[0081] In step S7, the raw material composite rendered image data is converted into a binary format that can be directly used by the device's dot matrix screen to obtain a raw material composite rendered image dot matrix frame. It is worth mentioning that the dot matrix screen is a device that displays images based on pixels and can only recognize and process binary data in a specific format. Converting the image data into a binary format suitable for the dot matrix screen is to ensure that the data can be correctly read and displayed by the device and to ensure compatibility between the hardware and the data. Different dot matrix screen devices may have different interface and data format requirements. Through this conversion, the raw material composite rendered image data can be matched with the specific device. The binary format is a format that can be directly processed and transmitted by a computer and is characterized by high efficiency and simplicity. After the image data is converted into the binary format, it can be quickly transmitted and processed within the device, reducing errors and delays during data transmission and improving the operating efficiency of the entire system. Compared with other complex image formats, the binary format is easier to parse and operate in the dot matrix screen's control system and can quickly transmit image data to each pixel for display. Therefore, by converting image data into a binary format that can be directly used by the dot matrix screen, we ensure that the color, brightness, and other information of each pixel are accurately transmitted to the dot matrix screen and displayed in the correct order and manner, thus presenting a complete and clear composite rendered image of the original material on the dot matrix screen. Only when the data format matches the device can the image be displayed correctly, avoiding problems such as garbled characters, distortion, or incomplete display.

[0082] In summary, the cloud platform-based dot matrix image data processing method based on the embodiment of the present application is explained. First, after the user uploads the frame image and real-time weather data, the cloud function is triggered to notify the cloud platform to generate a new task. Then the cloud platform receives the original material and performs image processing on the frame image to generate frame image data adapted to the dot matrix screen. At the same time, the cloud platform performs semantic understanding and image rendering on the real-time weather data to generate weather rendering image data. Subsequently, the cloud platform performs semantic collaborative interactive coding analysis based on visual features on the frame image data and the weather rendering image data to obtain synthetic rendering image data. Finally, the synthetic rendering image data is converted into a binary format that can be directly used by the dot matrix screen to obtain dot matrix frame data. In this way, the local resource performance bottleneck can be broken through, and at the same time, the accurate adaptation and real-time interactive presentation of the image on the dot matrix screen can be ensured.

[0083] Figure 6 FIG is a system block diagram of a cloud platform-based dot matrix image data processing system according to an embodiment of the present application. Figure 6 As shown, according to an embodiment of the present application, a cloud platform-based dot matrix image data processing system 100 includes: an original material acquisition module 110, which is used to acquire original materials uploaded by users, and the original materials include frame images and real-time weather data; an upload notification module 120, which is used to trigger a cloud function of a file upload success event to notify the cloud platform of a new task after the original material is successfully uploaded; a cloud platform material acquisition module 130, which is used by the cloud platform to acquire the original material; a frame image processing module 140, which is used by the cloud platform to perform image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted for the dot matrix screen; a weather data rendering module 150, which is used by the cloud platform to perform semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendered image data; an image synthesis module 160, which is used by the cloud platform to synthesize the frame image data and the weather rendered image data to obtain original material synthesized rendered image data; a data format conversion module 170, which is used to convert the original material synthesized rendered image data into a binary format that can be directly used by the dot matrix screen of the device to obtain an original material synthesized rendered image dot matrix frame.

[0084] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the above-mentioned cloud platform-based dot matrix image data processing system 100 have been described in detail above. Figures 1 to 5 The description of the cloud platform-based dot matrix image data processing method has been introduced in detail, and therefore, its repeated description will be omitted.

[0085] In summary, the cloud platform-based dot matrix image data processing system 100 according to the embodiment of the present application is explained. First, after the user uploads the frame image and real-time weather data, the cloud function is triggered to notify the cloud platform to generate a new task. Then, the cloud platform receives the original material and performs image processing on the frame image to generate frame image data adapted to the dot matrix screen. At the same time, the cloud platform performs semantic understanding and image rendering on the real-time weather data to generate weather rendering image data. Subsequently, the cloud platform performs semantic collaborative interactive coding analysis based on visual features on the frame image data and the weather rendering image data to obtain synthetic rendering image data. Finally, the synthetic rendering image data is converted into a binary format that can be directly used by the dot matrix screen to obtain dot matrix frame data. In this way, the local resource performance bottleneck can be broken through, while ensuring the accurate adaptation and real-time interactive presentation of the image on the dot matrix screen.

[0086] In other embodiments of the present application, a multifunctional mobile lighting device with an RGB dot matrix display is also provided. Specifically, the RGB dot matrix display functionality is implemented based on a layered architecture design, with the driver layer, connection layer, and application layer working in concert. The driver layer transmits LED control signals compliant with the WS2812 protocol via the MOSI pin of the SPI interface. It uses DMA technology to directly transfer SPI buffered data for efficient transmission, ensuring a stable LED refresh rate above 60Hz. It also integrates a key encoder driver module to convert physical key inputs into logical events and uses an ADC to collect battery voltage data to monitor the battery status. The connection layer establishes a mapping between the physical layout of the LED dot matrix and logical coordinates, encapsulating underlying operations into standardized APIs. Its UI graphics library provides coordinate-based graphics rendering, supporting geometric drawing, coordinate transformation, and dynamic special effects processing. The application layer implements user interface navigation, battery visualization, charging animation control, interactive application logic, and dynamic power management. The operation logic menu manages the interface hierarchy, the application gameplay module calls the graphics library to implement games and dynamic effects, and the power management module adjusts operating parameters based on the device status. Interaction between layers is achieved through data and event streams: the driver layer passes raw key events and power data to the connection layer. After coordinate calibration and API encapsulation, the application layer calls the graphics library to generate display data, which drives the LED dot matrix update via the SPI interface and DMA channel. The RGB light board driver process sequentially performs hardware initialization, LED coordinate calibration, user input capture, application logic processing, graphics rendering, and SPI data transmission, forming a closed-loop control from user interaction to hardware display. The UI graphics library is responsible for converting application layer commands into display data, while the key driver and SPI module respectively complete input event generation and signal output, ultimately realizing the multi-functional interactive dot matrix display function.

Claims

1. A method for processing dot matrix image data based on a cloud platform, characterized in that: include: Acquire original materials uploaded by users, the original materials including frame images and real-time weather data; After the original material is uploaded successfully, the file upload success event triggers a cloud function to notify the cloud platform of a new task; The cloud platform obtains the original material; The cloud platform performs image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen; The cloud platform performs semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data; The cloud platform synthesizes the frame image data and the weather rendering image data to obtain the original material synthesized rendering image data, including: performing semantic collaborative interaction encoding based on visual features on the frame image data and the weather rendering image data to obtain a frame image-weather rendering image visual feature interaction matrix; and obtaining the original material synthesized rendering image data based on the frame image-weather rendering image visual feature interaction matrix; The raw material composite rendered image data is converted into a binary format that can be directly used by a dot matrix screen of a device to obtain a raw material composite rendered image dot matrix frame.

2. The cloud platform-based dot matrix image data processing method according to claim 1, characterized in that: The cloud platform performs image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen, including: Scaling the frame image to the pixel size of the dot matrix screen; The scaled frame image is subjected to color quantization processing to adapt to the color capability of the dot matrix screen to obtain the frame image data.

3. The cloud platform-based dot matrix image data processing method according to claim 1, characterized in that: The cloud platform performs semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data, including: Performing semantic encoding on the real-time weather data to obtain a real-time weather semantic embedding encoding vector; The real-time weather semantic embedding coding vector is input into an image rendering engine based on a generative adversarial network model to obtain the weather rendering image data.

4. The cloud platform-based dot matrix image data processing method according to claim 1, characterized in that: Performing semantic collaborative interaction encoding based on visual features on the frame image data and the weather rendering image data to obtain a frame image-weather rendering image visual feature interaction matrix includes: Performing visual feature extraction on the frame image data to obtain a frame image visual feature encoding feature vector; Performing visual feature extraction on the weather rendering image data to obtain a weather rendering image visual feature encoding vector; The frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector are subjected to performance-intrinsic visual feature semantic collaborative encoding to obtain the frame image-weather rendering image visual feature interaction matrix.

5. The cloud platform-based dot matrix image data processing method according to claim 4, characterized in that: Performing visual feature extraction on the frame image data to obtain a frame image visual feature encoding feature vector includes: performing visual feature extraction based on a feature pyramid network on the frame image data to obtain the frame image visual feature encoding feature vector.

6. The cloud platform-based dot matrix image data processing method according to claim 4, characterized in that: Performing performance-intrinsic visual feature semantic collaborative encoding on the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector to obtain the frame image-weather rendering image visual feature interaction matrix, including: performing kernel coding based on feature principal components on the frame image visual feature coding feature vector and the weather rendering image visual feature coding vector to obtain a set of frame-weather rendering image visual feature principal component kernel coding vectors; Calculating a performance operator between any two frame-weather rendered image visual feature principal component kernel encoding vectors in the set of the frame-weather rendered image visual feature principal component kernel encoding vectors to obtain a frame-weather rendered image visual feature performance operator topology matrix; The set of principal component kernel encoding vectors of the frame-weather rendering image visual features and the frame-weather rendering image visual feature performance operator topology matrix are graph convolutionally encoded to obtain the frame image-weather rendering image visual feature interaction matrix.

7. The cloud platform-based dot matrix image data processing method according to claim 6, characterized in that: Performing kernel coding based on feature principal components on the frame image visual feature coding feature vector and the weather rendering image visual feature coding vector to obtain a set of frame-weather rendering image visual feature principal component kernel coding vectors, comprising: Performing principal component analysis on the frame image visual feature encoding feature vector and the weather rendering image visual feature encoding vector to obtain a set of frame image visual feature principal component encoding feature vectors and a set of weather rendering image visual feature principal component encoding vectors; Each corresponding group of frame image visual feature principal component encoding feature vectors and weather rendering image visual feature principal component encoding vectors in the set of the frame image visual feature principal component encoding feature vectors and the set of the weather rendering image visual feature principal component encoding vectors are respectively subjected to principal component kernel encoding to obtain the set of the frame-weather rendering image visual feature principal component kernel encoding vectors.

8. The cloud platform-based dot matrix image data processing method according to claim 7, characterized in that: The method further comprises performing graph convolution encoding on a set of principal component kernel encoding vectors of the frame-weather rendered image visual feature and a performance operator topology matrix of the frame-weather rendered image visual feature to obtain a frame image-weather rendered image visual feature interaction matrix, including: Based on the frame-weather rendered image visual feature performance operator topology matrix, performing cross-section compactification tuning based on dynamic gauge field coupling on each frame-weather rendered image visual feature principal component kernel encoding vector in the set of frame-weather rendered image visual feature principal component kernel encoding vectors to obtain a set of frame-weather rendered image visual feature principal component kernel optimized encoding vectors; The set of principal component kernel optimization coding vectors of the frame-weather rendering image visual feature and the frame-weather rendering image visual feature performance operator topology matrix are graph convolutionally coded to obtain the frame image-weather rendering image visual feature interaction matrix.

9. The cloud platform-based dot matrix image data processing method according to claim 8, characterized in that: Based on the frame image-weather rendering image visual feature interaction matrix, the original material synthetic rendering image data is obtained, including: inputting the frame image-weather rendering image visual feature interaction matrix into an image synthesis engine based on a diffusion model to obtain the original material synthetic rendering image data.

10. A cloud platform-based dot matrix image data processing system, characterized in that: include: The original material acquisition module is used to acquire the original material uploaded by the user, and the original material includes frame images and real-time weather data; The upload notification module is used to trigger a cloud function to notify the cloud platform of a new task when the original material is uploaded successfully. A cloud platform material acquisition module is used for the cloud platform to acquire the original material; A frame image processing module is used for the cloud platform to perform image processing on the frame images in the original material to obtain frame image data, and the frame image data is adapted to the dot matrix screen; A weather data rendering module is used on the cloud platform to perform semantic understanding and image rendering on the real-time weather data in the original material to obtain weather rendering image data; An image synthesis module is configured to synthesize the frame image data and the weather rendering image data on the cloud platform to obtain raw material synthesized rendering image data, wherein the image synthesis module is configured to: perform semantic collaborative interaction encoding based on visual features on the frame image data and the weather rendering image data to obtain a frame image-weather rendering image visual feature interaction matrix; and obtain the raw material synthesized rendering image data based on the frame image-weather rendering image visual feature interaction matrix; The data format conversion module is used to convert the raw material composite rendering image data into a binary format that can be directly used by the device's dot matrix screen to obtain the raw material composite rendering image dot matrix frame.

Images