Electronic Device AI Upscaling with Pre-Processing Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic devices face issues in providing high-resolution images with detailed content due to data reduction through downscaling and pre-processing, leading to loss of image details, especially in live streaming scenarios.
Innovation Solution
Implementing artificial intelligence models for interlocking encoding and decoding processes, utilizing neural networks to upscale images based on pre-processing information, including filter details and network status, to restore lost image details while reducing data and computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If downscaling and pre-processing are applied to reduce data amount, then the amount of data and computation is reduced, but image detail is lost
Solution Approach 1:
The transmitting device performs pre-processing (such as filtering) on the original image before downscaling and encoding. This preliminary action prepares the image in a way that preserves essential information while reducing data量, enabling the receiving device to reconstruct details more effectively during the upscaling process
Solution Approach 2:
The system transmits pre-processing related information (including filter types and parameters) from the transmitting device to the receiving device. This feedback mechanism allows the receiving device to understand what processing was applied and adjust its upscaling algorithm accordingly, thereby recovering image details that would otherwise be lost
2Device complexity
If a fixed artificial intelligence model is used for upscaling, then the upscaling process is simple, but it cannot adapt to different pre-processing conditions
Solution Approach 1:
The receiving device dynamically selects or configures the artificial intelligence upscaling model based on the received pre-processing related information. Instead of using a single fixed model, the system adjusts model parameters or selects different models according to the specific pre-processing conditions, achieving adaptability without excessive complexity
Solution Approach 2:
The system changes parameters of the artificial intelligence model based on pre-processing related information. By adjusting model parameters (such as filter coefficients, layer configurations, or processing intensity) according to the transmitted pre-processing conditions, the model adapts to different scenarios while maintaining a relatively simple structure
3Loss of information
If detailed pre-processing information is transmitted to restore image details, then image quality is improved, but data transmission amount increases
Solution Approach 1:
Instead of transmitting the entire pre-processed image or all possible pre-processing information, the system extracts only the essential pre-processing related information (such as filter type identifiers and key parameters). This extracted information is sufficient for the receiving device to reconstruct image details while keeping the transmitted data量 minimal
Solution Approach 2:
The pre-processing related information acts as an intermediary that bridges the transmitting and receiving devices. Rather than transmitting large amounts of image data or complex processing instructions, this compact intermediary information enables the receiving device to infer and apply the appropriate reconstruction algorithms, achieving detailed restoration with minimal data transmission
Data Source
AI summary
An example electronic device may include a memory configured to include at least one instruction; and a processor configured to be connected to the memory to control the electronic device, and obtain an output image by upscaling an input image using an artificial intelligence model trained to upscale an image, wherein the processor is configured to control the electronic device to: obtain parameter information of the artificial intelligence model based on pre-processing related information performed on the input image, and upscale the input image using the artificial intelligence model corresponding to the obtained parameter information.