Compressed Reference Frames in Video Transcoders
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video compression technologies rely on uncompressed reference frames, leading to high bandwidth and storage requirements, which increase system costs and can result in video artifacts due to errors propagating across frames.
Innovation Solution
A transcoder system that compresses reference frames by transforming pixel blocks from the spatial domain to the frequency domain using a Modified Hadamard Transform, allowing for storage in a compressed format, reducing bandwidth and storage needs while minimizing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If reference frames are stored uncompressed to preserve video quality, then video quality is maintained, but memory and bandwidth requirements increase significantly
Solution Approach 1:
The patent transforms reference frames from spatial domain to frequency domain using Hadamard transform, changing the representation parameters of the data. This transformation allows the same visual information to be stored in a compressed form while maintaining the ability to reconstruct high-quality images, thus resolving the contradiction between video quality and storage requirements
Solution Approach 2:
The patent creates compressed copies of reference frames in frequency domain instead of storing original spatial domain data. These compressed copies contain sufficient information for motion compensation and prediction, reducing memory and bandwidth requirements while maintaining video quality through efficient data representation
2Quantity of substance
If reference frames are compressed to reduce storage and bandwidth, then system cost decreases, but video artifacts may increase due to error propagation
Solution Approach 1:
By changing from spatial to frequency domain representation, the patent enables lossless or near-lossless compression of reference frames. The frequency domain transformation preserves important visual information while discarding less significant details, reducing storage requirements without introducing noticeable artifacts when properly decoded
Solution Approach 2:
The frequency domain acts as an intermediary representation between the original spatial domain reference frames and the compressed storage format. This intermediate transformation allows for efficient compression while maintaining the ability to reconstruct accurate reference frames for motion compensation, preventing error propagation
3Measurement precision
If uncompressed reference frames are used in transcoding, then decoding accuracy is maintained, but processing hardware complexity increases
Solution Approach 1:
The patent changes the domain parameter from spatial to frequency representation, allowing reference frames to be processed and stored in a more efficient format. The Hadamard transform provides a mathematically equivalent representation that requires less hardware resources to manipulate while maintaining decoding accuracy through reversible transformation
4Quantity of substance
If reference frames are compressed using traditional methods, then storage efficiency improves, but drift and artifacts propagate across frames
Solution Approach 1:
The patent applies frequency domain transformation (Hadamard transform) as a consistent compression method across all reference frames in the transcoding process. This uniform transformation approach ensures that all frames are compressed and decompressed using the same mathematical operations, preventing drift and maintaining consistency across the entire video sequence
Data Source
AI summary
A system (and a method) for compressing reference frames in a video transcoder. A transcoder receives a compressed input stream in a first compressed format and output a compressed output stream in a second compressed format. A decoder and an encoder in the transcoder use compressed reference frames. The reference frames are compressed by transforming a block of pixels from a spatial domain to a frequency domain to generate a coefficient array. The coefficient array is quantized and encoded to compress the size of the coefficients array to the size of a fixed bucket. The values of the entropy coded and quantized array are stored in a memory for use in decoding and/or encoding.