Automated feature detection and identification in digital point-ordered signals
A computer-based automated method to detect and identify features in digital point-ordered signals is disclosed. The method is used for processing of non-destructive test signals, such as eddy current signals obtained from calibration standards. The signals are first automatically processed to remove noise and to determine a baseline. Next, features are detected in the signals using mathematical morphology filters. Finally, verification of the features is made using an expert system of pattern recognition methods and geometric criteria. The method has the advantage that standard features can be, located without prior knowledge of the number or sequence of the features. Further advantages are that standard features can be differentiated from irrelevant signal features such as noise, and detected features are automatically verified by parameters extracted from the signals. The method proceeds fully automatically without initial operator set-up and without subjective operator feature judgment. 19 figs.
- Assignee:
- Dept. of Energy, Washington, DC (United States)
- Patent Number(s):
- US 5,737,445/A/
- Application Number:
- PAN: 8-425,758
- OSTI ID:
- 619395
- Resource Relation:
- Other Information: PBD: 7 Apr 1998
- Country of Publication:
- United States
- Language:
- English
Similar Records
Automated feature detection and identification in digital point-ordered signals
Method and apparatus for automatically detecting patterns in digital point-ordered signals