Remote Session Graphics Classification for Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for compressing graphical output in remote presentation sessions lack an efficient method to classify frames, leading to suboptimal encoding that can degrade user experience due to uneven compression quality across different graphic types.
Innovation Solution
A machine learning-based approach is used to classify graphics within frames, allowing for differential encoding of text and non-text elements, improving compression efficiency by utilizing a training set and test set to develop and verify the classification solution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If uniform compression is applied to all graphics in a frame, then bandwidth usage is reduced, but compression artifacts degrade user experience especially in text regions
Solution Approach 1:
The patent applies different compression quality levels to different regions of the frame based on their classification. Text regions are encoded with higher fidelity using text-optimized codecs, while non-text regions use standard compression. This local differentiation resolves the contradiction by maintaining high quality where needed (text) while still achieving overall bandwidth reduction through compression in other areas.
Solution Approach 2:
The patent segments the frame into multiple graphics or tiles and classifies each segment independently as text or non-text. This segmentation allows differential encoding where text segments receive higher quality encoding while non-text segments use standard compression, thereby reducing overall bandwidth usage without degrading text quality.
2Manufacturing precision
If higher fidelity encoding is used for all graphics, then user experience is improved, but bandwidth consumption increases
Solution Approach 1:
The patent applies higher fidelity encoding selectively only to text regions that require it for readability, while applying standard compression to non-text regions where high fidelity is less critical. This resolves the contradiction by optimizing the balance between quality and bandwidth consumption on a regional basis.
Solution Approach 2:
The patent changes encoding parameters dynamically based on graphic classification. Text graphics receive encoding parameters optimized for text fidelity (higher quality settings), while non-text graphics use different parameters optimized for compression efficiency. This parameter adaptation resolves the contradiction by matching encoding intensity to actual content requirements.
3Productivity
If machine learning classification is implemented, then differential encoding efficiency is improved, but system complexity increases
Solution Approach 1:
The patent implements a self-training mechanism where the machine learning classifier automatically learns from classified graphics data. The system uses a training set of labeled graphics to train the classifier, then evaluates it on a test set. This self-service approach to training reduces the need for manual configuration and complex external training pipelines, thereby improving compression efficiency while managing system complexity.
Data Source
AI summary
An invention is disclosed for classifying a graphic—e.g. as text or non-text. In embodiments, machine learning is used to generate a solution for classifying graphics of a graphic based on providing the machine learning system a plurality of graphics that are already classified. The way to determine a classification is then used by a remote presentation session server to classify tiles of frames to be transmitted to a client in a remote presentation session. The server encodes the tiles based on their classifications and transmits the encoded tiles to the client.