Image Compression Pre-Processing Network for Standard Codec Compatibility
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Standard image compression techniques struggle to enhance image quality while maintaining compatibility, as methods that improve compression rate often require modifying the compression standard algorithm, limiting their compatibility with existing standards.
Innovation Solution
An image compression apparatus and method that incorporates a pre-processing network and a codec modeling network to simulate standard compression techniques, along with a bit estimation network, to enhance image quality and compression rate without altering existing standards, using artificial neural networks for learning and error backpropagation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If prediction performance enhancement methods (adding novel prediction modes, using machine learning-based prediction) are used to reduce the amount of data to be encoded, then the compression rate is improved, but the compatibility with existing standard compression techniques deteriorates
Solution Approach 1:
The patent applies preliminary action by performing image preprocessing (denoising, sharpening, artifact removal) before the standard compression encoding process. This pre-processing prepares the image data in advance to be more amenable to standard compression algorithms, thereby improving compression efficiency without modifying the standard encoding process itself, thus maintaining compatibility.
2Manufacturing precision
If post-compression quality enhancement techniques (pixel adaptive noise reduction filter, CNN) are used to restore image quality, then the image quality is improved, but the compression rate deteriorates because the original image information is already lost during compression
Solution Approach 1:
The patent performs quality enhancement operations (denoising, sharpening, artifact removal) as pre-processing steps before compression encoding. By enhancing the image quality in advance, the subsequent compression can achieve better compression rates while maintaining high quality, avoiding the need for post-compression enhancement that would compromise compression efficiency.
Solution Approach 2:
The patent uses learning models trained on standard compression processes to create a simulated decoded image that copies the characteristics of actual standard compression output. This allows the system to learn and compensate for compression artifacts without actually performing the lossy compression during the enhancement phase, thereby preserving both quality and compression efficiency.
3Productivity
If stronger quantization is applied to minimize image data information, then the compression rate is improved, but the image quality deteriorates
Solution Approach 1:
The patent applies pre-processing operations (denoising, sharpening) before quantization and compression. These preliminary operations enhance the image quality and structure in advance, making the image more robust to subsequent strong quantization. This allows the system to apply stronger quantization for better compression while the pre-processed image maintains acceptable quality despite the information loss.
Data Source
AI summary
An image compression apparatus includes: an image acquisition unit configured to acquire a raw data image; a pre-processing network configured to receive the raw data image and pre-process the raw data image according to a pattern estimation method learned beforehand; and an encoder unit configured to receive the pre-processed image and compress the pre-processed image according to a pre-designated standard compression technique to output a compressed image. The pre-processing network, which can be added during learning and can be implemented as an artificial neural network, can have learned beforehand by way of a backpropagation of a restoration error through a codec modeling unit that has learned beforehand to simulate a standard codec unit, where the restoration error can be obtained by comparing a restored image obtained based on a simulated decoded image with the raw data image.