Neural Image Compression With Latent Coding for Low-Distortion Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is an increasing demand for higher resolution and lower distortion image and video content over communications networks, which strains network resources and energy consumption, necessitating improved compression methods that reduce data transmission without compromising image quality.
Innovation Solution
A computer-implemented method using trained neural networks for encoding and decoding images and videos, involving latent representation, quantization, and entropy encoding to transmit bitstreams, allowing for reduced output distortion or file size based on fixed parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If image and video content is transmitted in higher resolution with lower distortion, then image quality is improved, but network resource consumption and energy use increase
Solution Approach 1:
The patent applies preliminary action by performing image enhancement and restoration operations before transmission. The system pre-processes images to improve their quality, sharpness, and detail using neural networks and super-resolution algorithms, so that when images are compressed and transmitted at lower resolutions, they can still be reconstructed with high quality at the receiving end, thereby reducing network energy consumption while maintaining image quality
Solution Approach 2:
The patent replaces traditional mechanical compression and transmission systems with AI-based neural network systems. Instead of relying on conventional compression algorithms that lose information, the system uses deep learning models to learn efficient representations and reconstruct high-quality images from compressed data, substituting mathematical/computational processes for traditional signal processing mechanisms
2Loss of energy
If image compression is increased to reduce data transmission, then network resource use is reduced, but output image distortion increases
Solution Approach 1:
The patent introduces an intermediary reconstruction system at the receiving end that uses neural networks to restore and enhance compressed images. This intermediary process bridges the gap between compressed input and high-quality output, allowing aggressive compression to be applied during transmission while maintaining final image quality through AI-based restoration, thus reducing network energy use without sacrificing output image quality
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting compression levels, resolution, and quality parameters based on network conditions, device capabilities, and content characteristics. The system transforms images through multiple parameter transformations including super-resolution scaling, sharpness enhancement, and adaptive compression ratios, allowing optimization of the balance between transmission efficiency and output quality
Data Source
AI summary
A computer-implemented method for lossy image or video compression, transmission and decoding, the method including the steps of (i) receiving an input image at a first computer system; (ii) encoding the input image using a first trained neural network, using the first computer system, to produce a latent representation; (iii) quantizing the latent representation using the first computer system to produce a quantized latent; (iv) entropy encoding the quantized latent into a bitstream, using the first computer system; (v) transmitting the bitstream to a second computer system; (vi) the second computer system entropy decoding the bitstream to produce the quantized latent; (vii) the second computer system using a second trained neural network to produce an output image from the quantized latent, wherein the output image is an approximation of the input image.