Correlation Function Generation for Wave Source Direction Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques fail to generate a correlation function with a clear peak in environments with high ambient noise levels, making it difficult to accurately estimate the direction of sound or wave sources.
Innovation Solution
A correlation function generation apparatus and method that includes input signal obtaining units, a transformer for frequency domain signal conversion, a cross-spectrum calculator, a variance calculator, and a correlation function calculator, which calculate and weight the cross-spectrum based on variance to produce a clear peak correlation function even in noisy conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional cross-spectrum calculation methods are used in high noise environments, then the calculation process is simple, but the correlation function peak becomes unclear and direction estimation accuracy deteriorates
Solution Approach 1:
The patent segments the cross-spectrum calculation process by introducing intermediate steps: calculating variance of phase components, identifying reliable frequency components based on variance thresholds, and separately processing reliable and unreliable components. This segmentation allows the system to handle noise by focusing computational resources on reliable components while filtering out noise-dominated components, thereby improving direction estimation accuracy without uniformly increasing complexity across all frequency components.
Solution Approach 2:
The patent applies local quality by treating different frequency components differently based on their individual variance characteristics. Each frequency component is evaluated independently, and only those with variance below a threshold are considered reliable for direction estimation. This local evaluation ensures that noise-affected frequency components do not degrade the overall correlation function quality, while maintaining simple processing for components that are already reliable.
2Reliability
If variance-based weighting is applied to frequency components, then the correlation function peak clarity improves in noisy environments, but the calculation time and processing complexity increase
Solution Approach 1:
The patent implements partial action by selectively applying variance calculation and weighting only to frequency components where it is necessary. Instead of processing all frequency components uniformly, the system identifies components with high variance (indicating noise dominance) and applies sophisticated weighting only where needed. This approach achieves sufficient correlation function peak clarity in noisy environments while avoiding unnecessary computational overhead in clean frequency regions.
Solution Approach 2:
The patent changes the parameter of frequency component reliability by introducing variance as a dynamic parameter. The variance value determines the weight assigned to each frequency component, allowing the system to adaptively adjust the influence of each component based on current noise conditions. This parameter-based approach enables the correlation function to maintain clear peaks under varying noise levels while keeping the processing framework flexible and efficient.
Data Source
AI summary
A correlation function having a clear peak is generated even in an environment in which an ambient noise level is high. A correlation function generation apparatus (100) includes a plurality of input signal obtaining units (101), a changing unit (102), a cross-spectrum calculator (103), a variance calculator (104), and a correlation function calculator (105). The input signal obtaining unit (101) obtains a wave generated by a wave source as an input signal. The transformer (102) obtains a plurality of frequency domain signals by transforming a plurality of input signals obtained by the plurality of input signal obtaining units. The cross-spectrum calculator (103) calculates a cross-spectrum based on the plurality of frequency domain signals. The variance calculator (104) calculates the variance of the cross-spectrum. The correlation function calculator (105) calculates and generates a correlation function based on the cross-spectrum and the variance.