Objective Difference Grade

Description

The Objective Difference Grade (ODG) is a key metric defined within the 3GPP specifications for the objective perceptual evaluation of video quality. It operates by algorithmically comparing a processed or degraded video sequence against its original, pristine reference. The core methodology involves extracting perceptual features from both video streams, such as spatial and temporal information, contrast, and edge details. These features are then analyzed to model human visual perception, predicting the subjective quality score a typical viewer would assign. The output is a single numerical score, typically ranging from 0 (imperceptible impairment) to negative values indicating increasing levels of perceived degradation (e.g., -1 for perceptible but not annoying, -2 for slightly annoying, -3 for annoying, -4 for very annoying). This automated process eliminates the need for costly and time-consuming subjective testing panels for routine quality checks.

Architecturally, ODG calculation is often integrated into test equipment, network probes, or software analysis tools used by operators and equipment vendors. It relies on complex perceptual models, such as those standardized by the Video Quality Experts Group (VQEG) or ITU-T, which are referenced or adapted within 3GPP specs. The metric works by processing video frames through a model that simulates the human visual system's sensitivity to various types of distortions, including compression artifacts (blocking, blurring), packet loss (freezing, slicing), and transmission errors. Key components of the evaluation include a precise temporal alignment of the reference and test sequences, color space conversion to a perceptually uniform space (like CIELAB), and multi-scale spatial analysis to account for viewing distance and display characteristics.

Its role in the 3GPP ecosystem is primarily in the specification and verification of video codecs and streaming services. For instance, when a new video codec profile is standardized, its performance is evaluated not just by bitrate savings but also by the ODG scores it achieves at various bitrates compared to a reference codec. This ensures that compression efficiency gains do not come at an unacceptable cost to perceptual quality. Network operators use ODG-based monitoring to assure end-user QoE for services like Mobile TV, video calling, and streaming. By providing an objective, repeatable measure, ODG enables consistent benchmarking across different vendors' implementations, network conditions, and device types, forming a critical part of the end-to-end quality management chain for multimedia services in mobile networks.

Purpose & Motivation

The ODG was introduced to address the critical need for an efficient, scalable, and standardized method to evaluate video quality in mobile networks. Prior to its adoption, quality assurance heavily relied on subjective testing methods like Mean Opinion Score (MOS), where human viewers rate video sequences. While accurate, this approach is prohibitively expensive, slow, and not reproducible at scale for the vast array of codecs, devices, and network conditions in a live mobile ecosystem. The industry needed an objective correlate to subjective opinion that could be automated for continuous testing, codec development, and network optimization.

The creation of ODG was motivated by the explosive growth of mobile video traffic and the introduction of advanced video codecs like H.264/AVC and later HEVC. As operators competed on service quality, a standardized metric was necessary to specify minimum quality requirements in technical specifications and service level agreements (SLAs). It solves the problem of quantifying the perceptual impact of compression and transmission impairments in a way that aligns with human experience. This allows engineers to make informed trade-offs between bandwidth usage and visual quality, and to proactively detect quality degradation in the network before it impacts a large number of subscribers, thereby improving customer satisfaction and reducing churn.

Key Features

• Provides a standardized objective score correlating to subjective human perception (e.g., MOS).
• Enables automated, high-volume video quality testing without human panels.
• Used for benchmarking and performance verification of video codecs (e.g., AVC, HEVC, VVC).
• Supports quality monitoring and assurance for mobile video streaming and broadcast services.
• Integrates perceptual models accounting for spatial, temporal, and color vision characteristics.
• Delivers a single metric (e.g., on a scale from 0 to -4) for easy interpretation and alerting.

Evolution Across Releases

Defining Specifications