Encoding and decoding video data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video streaming technologies, particularly for mobile devices and game streaming, face challenges with bandwidth usage due to the need for frequent I-frames, leading to high bandwidth requirements and decoding issues on devices with limited computational power and unstable connections.
Innovation Solution
Creating a 'dictionary' dataset of images at the client device, which allows for encoding video data by replacing full frames with metadata, reducing the need for transmitting I-frames and optimizing bandwidth usage without modifying hardware or codec standards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If frequent I-frames are transmitted to ensure decoding reliability, then decoding reliability is improved, but bandwidth consumption increases
Solution Approach 1:
The client device pre-loads a dictionary of candidate frame data before video streaming begins. This preliminary action enables the decoder to reconstruct I-frames locally without receiving them over the network, eliminating the need to transmit frequent I-frames while maintaining decoding reliability.
Solution Approach 2:
The invention extracts the I-frame data from the bitstream and replaces it with metadata identifiers that point to pre-loaded dictionary entries on the client device. This extraction removes the bandwidth-consuming I-frame transmissions while preserving the essential reference data locally.
2Reliability
If more I-frames are transmitted to handle poor connections, then decoding reliability is improved, but bandwidth usage increases
Solution Approach 1:
The client device creates a local copy of essential frame data in the dictionary before streaming begins. This copying allows the device to reconstruct I-frames from local storage rather than receiving them over the network, maintaining reliability during poor connections without increasing bandwidth usage.
3Manufacturing precision
If video streaming uses standard GOP structures with I-frames, then video quality is maintained, but bandwidth requirements increase
Solution Approach 1:
The dictionary of candidate frames is pre-loaded into client memory before streaming begins. This preliminary action allows P-frames to reference locally-stored data instead of requiring I-frame transmissions, maintaining video quality while reducing bandwidth requirements.
Solution Approach 2:
The invention extracts I-frames from the traditional GOP structure and replaces them with metadata identifiers. This extraction maintains video quality by preserving reference frame data locally while eliminating the bandwidth overhead of transmitting complete I-frames.
4Manufacturing precision
If game streaming uses small GOPs with frequent I-frames, then video quality is maintained, but bandwidth requirements become extremely high
Solution Approach 1:
The client device pre-loads a dictionary of game-specific candidate frames into local memory. This copying enables the decoder to reconstruct frequent I-frames required by small GOP structures from local storage, maintaining high video quality for game streaming without the extremely high bandwidth requirements of traditional approaches.
Data Source
AI summary
A method, computing device and a computer-readable storage medium for encoding and decoding video data for streaming are provided. The method includes receiving video data comprising frames; accessing a dataset comprising at least one image expected to have similarity to at least one of the frames of the received video data; selecting an image or a portion of the image, from the dataset based on similarity to at least a portion of a first frame of the received video data; generating a frame using the image or the portion of the image selected from the dataset; encoding the generated frame and the frames of the received video data to produce encoded video data; and removing data corresponding to the generated frame from the encoded video data to produce modified encoded video data and generating data identifying the image or the portion of the image selected from the dataset.