Real time continuous wavelet transform implementation on a DSP processor.
J Med Eng Technol
; 33(3): 223-31, 2009.
Article
in En
| MEDLINE
| ID: mdl-19340693
ABSTRACT
The continuous wavelet transform (CWT) is an effective tool when the emphasis is on the analysis of non-stationary signals and on localization and characterization of singularities in signals. We have used the B-spline based CWT, the Lipschitz Exponent (LE) and measures derived from it to detect and quantify the singularity characteristics of biomedical signals. In this article, a real-time implementation of a B-spline based CWT on a digital signal processor is presented, with the aim of providing quantitative information about the signal to a clinician as it is being recorded. A recursive algorithm implementation was shown to be too slow for real-time implementation so a parallel algorithm was considered. The use of a parallel algorithm involves redundancy in calculations at the boundary points. An optimization of numerical computation to remove redundancy in calculation was carried out. A formula has been derived to give an exact operation count for any integer scale m and any B-spline of order n (for the case where n is odd) to calculate the CWT for both the original and the optimized parallel methods. Experimental results show that the optimized method is 20-28% faster than the original method. As an example of applying this optimized method, a real-time implementation of the CWT with LE postprocessing has been achieved for an EMG Interference Pattern signal sampled at 50 kHz.
Full text:
1
Collection:
01-internacional
Database:
MEDLINE
Main subject:
Algorithms
/
Signal Processing, Computer-Assisted
/
Electromyography
Type of study:
Prognostic_studies
Limits:
Humans
Language:
En
Journal:
J Med Eng Technol
Year:
2009
Document type:
Article
Affiliation country: