Sensor Signal Interval Filtering With Sharp Edge Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing discrete-time filtering methods for sensor signals often compromise between noise reduction and signal latency, with conventional methods either reducing noise at the cost of higher latency or achieving suboptimal noise reduction and edge steepness.
Innovation Solution
A method where the specific signal value is compared with a value interval spanned by two further signal values, and either retained or replaced by a new value determined based on this comparison, to minimize noise and interference while maintaining high edge steepness, using a cascaded filtering approach.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If complex filtering methods (e.g., frequency transformation, overlap-add method) are used to reduce noise interference, then noise reduction is improved, but signal latency increases
Solution Approach 1:
The patent divides the filtering process into multiple discrete stages, where each stage processes a specific time window of the input signal independently. This segmentation allows the system to apply filtering operations locally in time, avoiding the need to wait for complete signal blocks as required by frequency-domain methods, thereby reducing signal latency while maintaining noise reduction effectiveness.
Solution Approach 2:
The patent replaces complex frequency-domain filtering operations with time-domain arithmetic operations. Instead of performing Fourier transforms and frequency-domain manipulations that require processing entire signal blocks, the invention uses direct time-domain calculations with predetermined coefficients to achieve noise reduction with minimal delay.
2Object-affected harmful factors
If discrete-time filters (FIR or IIR) are used to filter sensor signals, then noise reduction is achieved, but signal latency is generated
Solution Approach 1:
The patent applies filtering operations selectively to specific portions of the signal rather than uniformly processing the entire signal through traditional filter structures. By applying filtering only where and when needed using local time-window operations, the system achieves noise reduction without the cumulative latency inherent in cascaded FIR or IIR filter stages.
3Object-affected harmful factors
If filtering methods are used to reduce noise, then noise reduction is improved, but edge steepness is reduced
Solution Approach 1:
The patent applies different processing characteristics to different regions of the signal. In time windows where edges or transitions are detected, the filtering is adjusted or bypassed to preserve edge steepness, while in steady-state regions, full filtering is applied to maximize noise reduction. This local adaptation allows simultaneous preservation of edge characteristics and noise reduction.
Data Source
Figure 1~3
Figure 4~5
Figure 6a~6b
AI summary
A method for filtering a sensor signal is proposed, comprising several steps. A discrete-time sensor signal is provided as an input signal. An output signal is generated using a continuous time-discrete filter on the input signal, which performs the following filtering steps for each value of the input signal: comparing a specific signal value of a specific time index of the input signal with two other signal values of the input signal, where the first of the two additional signal values precedes the specific signal value and the second of the additional signal values follows the specific signal value; and replacing the specific signal value in the output signal with a new value depending on a value interval spanned by the two additional signal values.