Bit-Depth Range SEI for Lower-Bit-Depth Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies struggle to efficiently encode high bit-depth content using lower bit-depth encoders and decoders, leading to compression errors, especially in the most significant bits, which are critical for accurate representation of depth maps and other high dynamic range content.
Innovation Solution
The proposed method allocates bits of the original source video into multiple bit depth ranges, coding each range separately within a single picture using subpictures or constituent rectangles, and signals a bit offset to position significant bits in higher order positions, allowing for lossy coding without requiring lossless encoding, and includes an SEI message for reconstructing the target picture with the same bit-depth as the original.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If high bit-depth content is encoded using lower bit-depth encoders, then device complexity is reduced, but compression errors increase especially in significant bits
Solution Approach 1:
The patent divides the bit-depth range into multiple segments (e.g., 8-bit and 12-bit ranges within a 16-bit depth image). Different segments are encoded separately using appropriate bit-depth encoders, allowing lower bit-depth devices to handle specific ranges while maintaining overall quality. This segmentation resolves the contradiction by enabling flexible encoding without requiring full high bit-depth capability across the entire image.
Solution Approach 2:
The patent introduces a new dimension of bit-offset information that tracks the position of significant bits within the bit-depth range. By adding this metadata dimension, the system can reconstruct the original high bit-depth image from lower bit-depth encoded segments, effectively compensating for compression errors through dimensional enrichment rather than raw bit-depth increase.
2Reliability
If bits are allocated to higher bit positions, then compression error impact is reduced, but bit allocation complexity increases
Solution Approach 1:
The patent performs preliminary bit allocation by pre-calculating and signaling the bit-offset values before encoding. This preliminary action establishes a clear mapping between source bits and coded bit positions, allowing the decoder to reconstruct the original bit sequence accurately without complex real-time calculations, thus reducing overall system complexity while maintaining error resilience.
Solution Approach 2:
The patent introduces bit-offset information as an intermediary element that mediates between the lower bit-depth encoded data and the original high bit-depth content. This intermediary metadata layer simplifies the relationship between encoding and decoding operations, making the system more manageable while ensuring that significant bits are positioned optimally to minimize compression error impact.
3Adaptability or versatility
If separate coding is performed for different bit ranges, then encoding flexibility is improved, but processing time increases
Solution Approach 1:
The patent merges multiple bit-range coding operations into a unified process where different bit-depth ranges are encoded in parallel within the same coding framework. By combining the processing of multiple ranges rather than handling them as completely separate operations, the system achieves encoding flexibility for different bit ranges while reducing overall processing time through parallelization and integrated resource utilization.
Data Source
AI summary
An apparatus including: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: allocate a first set of at least one bit of original content of a source picture to a first range of at least one bit; allocate a second set of at least one bit of the original content of the source picture to a second range of at least one bit; form a formed picture comprising a first region based on the first range of at least one bit and a second region based on the second range of at least one bit; code the formed picture into a coded picture.