HDR Video Encoder Contrast Analysis Bandwidth Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High Dynamic Range (HDR) video compression algorithms struggle to preserve visual quality by avoiding artifacts like contouring and banding while maintaining a reasonable bit rate, as current codecs fail to handle noisy signals effectively.
Innovation Solution
An encoder computes a residual image at full bit depth and performs contrast analysis to identify potential problematic areas, allocating more bandwidth and inserting high-frequency components to preserve accuracy, ensuring that the quantized image maintains visual quality during compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dithering is used to generate noise to avoid visual artifacts, then visual quality is improved, but bit rate increases far beyond admissible limits
Solution Approach 1:
The encoder performs contrast analysis to identify problematic areas (low-contrast gradients) and applies different quantization strategies locally. In identified problematic areas, the encoder maintains higher precision by using larger quantization steps or inserting high-frequency components, while in other areas standard quantization is used, thus avoiding global bit rate increase.
Solution Approach 2:
The encoder performs contrast analysis on the original image before compression to pre-identify potential problematic areas. Based on this analysis, the encoder pre-computes a plurality of quantizers tailored to different regions, allocating more bandwidth specifically to areas prone to contouring and banding artifacts, rather than uniformly increasing bit rate across the entire image.
2Adaptability or versatility
If contrast steps between consecutive coded levels are increased to cover larger luminance range, then HDR coverage is improved, but visual artifacts such as contouring and banding appear
Solution Approach 1:
The encoder applies different quantization precision to different regions of the image. In low-contrast gradient areas identified by contrast analysis, the encoder uses finer quantization steps or inserts high-frequency components to maintain smooth transitions, while in other areas larger contrast steps are used to efficiently cover the HDR luminance range.
Solution Approach 2:
The encoder dynamically adjusts quantization parameters based on local image characteristics. By computing multiple quantizers and selecting appropriate ones for different regions, the encoder adapts the quantization step size to local contrast levels, thereby maintaining both HDR coverage and visual quality.
Data Source
AI summary
A method for processing High Dynamic Range (HDR) video in order to improve the perceived visual quality of encoded content. An encoder receives an original image, wherein a contrast analysis is performed on said original image in order to highlight potential problematic areas. The original image is reduced to a decreased bit depth image in order to compute a predicted image, wherein the predicted image is then magnified to an increased bit depth image. A residual image is then computed from the original image and the increased bit depth image, and the residual image is transformed into a frequency signal. Quantized coefficients determined through the contrast analysis are then applied to the frequency signal to produce a reduced frequency signal. The encoder then inserts high-frequency components into the reduced frequency signal before encoding a compressed image from the reduced frequency image and the predicted image.