Embedded Lightweight Neural Network for Wide-Gamut Color Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems fail to accurately and efficiently restore wide-gamut color values in digital images due to gamut reduction, leading to inaccurate restorations and inefficient use of computational resources.
Innovation Solution
A gamut recovery system utilizing a lightweight neural network, specifically a multi-layer perceptron (MLP), is trained to predict wide-gamut color values by identifying out-of-gamut pixel values and their spatial coordinates, and the MLP is embedded within the smaller gamut digital image to facilitate accurate restoration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional systems reduce the range of visible colors to standard RGB spaces via clipping, then file size is optimized and display requirements are met, but color accuracy is lost and cannot be recovered
Solution Approach 1:
The system performs preliminary action by training a neural network model during the gamut reduction phase to predict wide-gamut color values. The model parameters are embedded in the image metadata before compression, enabling later recovery of accurate color information without increasing final file size.
Solution Approach 2:
A neural network model serves as an intermediary between the clipped RGB values and the original wide-gamut colors. The model predicts the lost color information by taking clipped pixel values as input and generating restored wide-gamut color values, bridging the gap caused by gamut reduction.
2Adaptability or versatility
If conventional systems use standard RGB color spaces, then display compatibility is improved, but wide-gamut color information is lost
Solution Approach 1:
The system embeds the neural network model parameters (a smaller structure) within the image metadata (a larger structure). This nested approach allows the compact model to be stored alongside the image data, enabling on-demand color restoration without adding significant overhead to the overall file size.
Solution Approach 2:
The system changes parameters by using a lightweight neural network model with optimized architecture (e.g., reduced layers, pruning, quantization) that can run efficiently on consumer devices. This parameter optimization enables the model to predict wide-gamut colors while maintaining fast processing speeds and low memory requirements.
3Measurement precision
If complex neural networks are used for gamut recovery, then color restoration accuracy is improved, but computational resource requirements increase
Solution Approach 1:
The system uses a disposable, lightweight neural network model that can be embedded in each image and executed locally without requiring persistent, complex infrastructure. The model is designed to be computationally inexpensive, allowing it to be applied to each image independently with minimal processing power and memory.
Solution Approach 2:
The neural network model is segmented into a compact architecture with fewer layers and parameters compared to traditional deep learning models. This segmentation reduces the computational burden while maintaining sufficient accuracy for gamut recovery, making it suitable for deployment on consumer devices with limited resources.
Data Source
AI summary
Systems, methods, and non-transitory computer-readable media embed a trained neural network within a digital image. For instance, in one or more embodiments, the systems identify out-of-gamut pixel values of a digital image in a first gamut, where the digital image is converted to the first gamut from a second gamut. Furthermore, the systems determine target pixel values of a target version of the digital image in the first gamut that correspond to the out-of-gamut pixel values. The systems train a neural network to predict the target pixel values in the first gamut based on the out-of-gamut pixel values. The systems embed the neural network within the digital image in the second gamut to allow for extraction of the embedded neural network from the digital image to restore the digital image to a larger gamut digital image.