Temporal Alignment of Audio Video Signals Using Characteristic Curves
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current objective measurement techniques for evaluating the quality of encoded audio or video signals face challenges in accurately temporally aligning disturbed and undisturbed signals, especially in complex scenarios like mobile radio communication or internet telephony, leading to underestimation of video quality and high computational costs.
Innovation Solution
The approach involves decomposing information signals into subunits, characterizing them with multiple characteristics, and using cross-correlation to align search patterns within characteristic vectors, reducing complexity and increasing reliability by examining characteristics rather than signal streams directly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If direct correlation of two signals is used for temporal alignment, then the method is simple to implement, but it fails to reliably align complex signals such as video signals under high-level disturbances
Solution Approach 1:
The patent introduces characteristic vectors as an intermediary representation between the original signal and the alignment process. Instead of directly correlating complex signal streams, the method extracts characteristic vectors that capture essential signal properties, making the alignment process more reliable while managing complexity through this intermediate abstraction layer
Solution Approach 2:
The patent transforms the alignment problem from operating on raw signal data to operating on extracted characteristic parameters. By changing the domain from time-domain signal values to feature-domain characteristic vectors, the method achieves better reliability for complex signals while the parameter extraction process manages the computational complexity
2Reliability
If simple alignment techniques are used for speech signals, then the computational effort is low, but the method cannot be reliably applied to complex signals such as video signals
Solution Approach 1:
The patent segments the complex signal processing task into distinct stages: characteristic extraction, characteristic vector formation, and alignment based on characteristic correlation. This segmentation allows the method to handle complex signals reliably by breaking down the problematic full-signal correlation into manageable steps that focus on essential features
Solution Approach 2:
The patent extracts characteristic vectors from the full signal streams, taking out only the essential features needed for alignment. This extraction process removes unnecessary computational burden from the alignment operation while preserving the reliability needed for complex signals by focusing on discriminative characteristics
3Measurement precision
If full signal correlation is performed for temporal alignment, then accurate alignment can be achieved, but the computational time becomes extremely large
Solution Approach 1:
The patent extracts characteristic vectors that capture the essential temporal structure of signals without requiring processing of the entire signal data. This extraction maintains alignment precision by preserving key temporal relationships in a compressed representation, dramatically reducing computational time compared to full signal correlation
Solution Approach 2:
The patent changes the operational parameters from using full signal data to using extracted characteristic parameters. This parameter transformation maintains measurement precision for temporal alignment while reducing computational time by operating on a smaller, more efficient parameter space
Data Source
AI summary
An apparatus for determining information in order to temporally align first and second information signals, each including one or several information values for successive sampling times, the apparatus including a characteristic extractor for extracting a characteristic per sampling time or per subsequence of sampling times from the first information signal within a first search pattern to obtain a first characteristic curve, and for extracting the characteristic per sampling time or per subsequence of sampling times from the second information signal within a second search pattern to obtain a second characteristic curve and a determiner for determining the information for temporal alignment based on similarities between the first and second characteristic curves of the characteristic by means of a first search for a search pattern from one of the first and second characteristic curves of the characteristic within the other of the first and second characteristic curves of the characteristic.