Video Range Reduction for Compression Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression techniques face limitations in achieving further compression beyond certain quality degradation thresholds, leading to visible distortions and bitrate constraints, especially with large quantization step sizes and frame dropping.
Innovation Solution
The technique involves preprocessing video data by scaling pixel samples in the spatial domain, allowing for additional compression through range reduction and matching reference frame ranges, which are then signaled and applied during decoding to enhance compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If lossy compression techniques are used to achieve dramatic bitrate decreases, then compression ratio is improved, but visible distortions such as blocking and ringing occur
Solution Approach 1:
The patent applies range reduction as a preliminary action before transform and quantization. By scaling pixel values to a reduced range (e.g., 0-255 to 0-127) before compression, the encoder prepares the data in a form that allows more aggressive quantization with fewer visible artifacts. This preliminary scaling modifies the input characteristics to enable better subsequent compression steps.
Solution Approach 2:
The patent changes the parameter range of pixel values through scaling operations. By transforming pixel values from the full range to a reduced range and then restoring them after decompression, the system effectively modifies the dynamic range parameter to achieve better compression characteristics. This parameter transformation allows the quantizer to operate more efficiently.
2Loss of energy
If quantization step size is increased to achieve further compression, then bitrate is reduced, but quantization artifacts and visible distortions increase
Solution Approach 1:
Range reduction is applied as a preliminary step before quantization. By scaling the pixel range down beforehand, the subsequent quantization operates on a compressed value space, allowing larger quantization steps to be applied without producing proportionally larger artifacts in the final image. The preliminary scaling effectively pre-compresses the value distribution.
Solution Approach 2:
The patent introduces an additional processing dimension by applying range reduction scaling before the standard compression pipeline. This extra transformation step in the value space creates a intermediate representation that bridges the full-range input and the quantized output, allowing more flexible control over the compression-accuracy tradeoff.
3Loss of energy
If frame dropping is used to reduce bitrate, then transmission bandwidth is reduced, but video quality and temporal continuity deteriorate
Solution Approach 1:
The patent changes the spatial domain parameter range of pixel values within each frame through range reduction. By modifying the value distribution and dynamic range of the video data, the technique enables better compression efficiency without requiring frame drops, thus maintaining temporal continuity while reducing bitrate through improved per-frame compression.
4Productivity
If resolution is reduced to decrease bitrate, then compression efficiency is improved, but image detail and quality suffer
Solution Approach 1:
The patent changes the value range parameter of pixels rather than reducing spatial resolution. By applying range reduction that scales pixel values (e.g., dividing by 2 and adjusting offset), the technique achieves compression efficiency improvements while preserving the original spatial dimensions and image details, avoiding the quality loss associated with resolution reduction.
Data Source
AI summary
Techniques and tools are described for decoding video data having samples that have been scaled in the spatial domain. For example, a decoder receives a bit stream that includes coded video data for a current frame. The decoder processes at least one syntax element (e.g., sequence layer flag, frame layer flag) that indicates whether the current frame should be scaled up in value in a spatial domain. If so, then the samples for the current frame are scaled up in value in the spatial domain. As another example, for a reference frame used in motion compensation for a current frame, a decoder scales samples of the reference frame so the range of the reference frame matches the range of the current frame.