Minus One Encoding for Video Reference Wrap-Around Offset
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards, such as ITU-T H.264 and H.265, face inefficiencies in reference offset signaling, particularly with the reference wrap-around offset being restricted to non-zero values and not allowing zero, which wastes bits and limits flexibility in signaling horizontal wrap-around motion compensation.
Innovation Solution
The proposed solution involves signaling a reference wrap-around offset using a minus one encoding scheme, allowing zero as a valid value and enabling flexible signaling of the offset in units of minimum coding block size, thereby improving encoding efficiency and bitstream compression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the reference wrap-around offset is restricted to non-zero values only, then the signaling mechanism avoids ambiguity in motion compensation, but it wastes bits and reduces flexibility in encoding efficiency
Solution Approach 1:
The patent inverts the traditional encoding approach by using a minus-one encoding scheme where the offset value is encoded as (offset - 1). This allows zero to be represented as -1 in the encoded form, enabling the offset to take zero as a valid value while maintaining unique decoding. The inversion transforms the restriction from 'non-zero offset values' to 'encoded values starting from -1', resolving the contradiction between signaling clarity and encoding efficiency.
Solution Approach 2:
The patent changes the parameter range of the reference wrap-around offset from strictly positive integers to non-negative integers (including zero). By modifying the domain of the offset parameter and adjusting the encoding scheme accordingly (using minus-one encoding), the system gains flexibility to represent no wrap-around offset (zero) while maintaining unambiguous decoding through the transformed parameter space.
2Reliability
If the reference wrap-around offset excludes zero, then the motion compensation avoids edge artifacts, but it limits the flexibility in handling different coding block sizes and picture widths
Solution Approach 1:
The patent modifies the parameter domain of the reference wrap-around offset to include zero, and adjusts the encoding scheme to accommodate this change. The minus-one encoding transforms the parameter space so that zero offset (no wrap-around) is represented as -1 in the bitstream, while positive offsets are represented as offset-1. This parameter transformation enables flexible adaptation to different coding block sizes and picture widths, including cases where no wrap-around is needed.
Solution Approach 2:
The patent introduces dynamic adaptability by allowing the reference wrap-around offset to be zero or positive based on the actual content and coding conditions. The encoding scheme dynamically adjusts to represent both zero and positive offset values, enabling the system to adapt to different picture widths, coding block sizes, and motion patterns without being constrained by a fixed non-zero requirement.
3Measurement precision
If more bits are allocated for offset signaling, then the precision of motion compensation improves, but the bitstream size increases reducing compression efficiency
Solution Approach 1:
The patent transforms the offset parameter representation using minus-one encoding, which shifts the value range. This parameter transformation allows the system to maintain precision in representing offset values (including zero) while optimizing the bit allocation. The transformed parameter space enables more efficient entropy coding by creating a distribution that better matches typical offset value frequencies, thereby reducing the average bits required per offset signal.
Solution Approach 2:
The patent discards the traditional constraint that offset must be non-zero, and recovers the ability to represent zero offset efficiently. By allowing zero and using the minus-one encoding scheme, the system can represent the most common case (no wrap-around) with a compact encoding, while still maintaining the ability to represent larger offsets when needed, thus optimizing the balance between precision and bitstream size.
Data Source
AI summary
Some techniques and apparatuses described herein provide generation of a set of orthogonal sequences for transmission of a signal including a payload using an orthogonal base sequence in a time domain. In one example, the orthogonal base sequence in the time domain may be a pi over 2 (pi/2) binary phase shift keying (BPSK) sequence in the time domain, such as prior to transform precoding for transmission. Some techniques and apparatuses described herein provide for the set of orthogonal sequences to be generated such that the set of orthogonal sequences are orthogonal within a symbol (e.g., an orthogonal frequency division multiplexing symbol) by applying an intra-symbol orthogonal cover code (OCC). Applying the intra-symbol OCC provides intra-symbol orthogonality. Thus, a peak to average power ratio of a user equipment is reduced for uplink transmissions and intra-symbol orthogonality is preserved.