| 摘要 |
The present invention addresses the problem of enhancing peak detection accuracy while making negative peak detection using a continuous wavelet transform possible. A continuous wavelet transform with varying scale factors is carried out on a signal waveform under analysis and wavelet coefficients for each time are calculated (S2). A ridge is detected on an image plotting the wavelet coefficient intensity against the scale factor and time, and positive and negative peak candidates are extracted on the basis of the ridge (S3, S4). Thereafter, peak position displacement and width displacement resulting from proximate peak influence is corrected (S5). Then, false negative peaks resulting from negative peak artifacts are removed through identification of peak shape asymmetry, etc. (S6). Subsequently, false peak groups resulting from the removal of high-frequency noise components, etc., and true peak groups, etc., are identified, and the results of the identification are used to remove false peaks (S7, S8). |
