Fig. 1. Normal FBG spectrum and various types of abnormal FBG spectra. (a) Normal spectrum; (b) Spectrum with Gaussian noise; (c) Spectrum with bad points; (d) Spectrum with baseline drift

Fig. 2. Different filter templates' filtering effects. (a) Original spectrum signal; (b) Gaussian function; (c) First-order derivative of the Gaussian; (d) Negative of the second-order derivative of the Gaussian

Fig. 3. One-dimensional zero-mean negative Gaussian second derivative

Fig. 4. Comparison of filtering performance of five algorithms under different amplitude noise

Fig. 5. Comparison of performance of five algorithms under different amplitude noises

Fig. 6. Comparison of the filtering effect simulated spectra with bad points

Fig. 7. Comparison of filtering performance of five algorithms under different total number of bad points

Fig. 8. Comparison of filtering effects of simulated spectra with baseline

Fig. 9. Comparison of the performance of five filtering algorithms under different baseline orders

Fig. 10. Comparison of filtering performance of five algorithms. (a) Spectral signals with baseline, Gaussian noise, and bad points; (b) The signals by sliding correlation filtering method; (c) The signals by Gaussian filtering method; (d) The signals by median mean filtering method; (e) The signals by median filtering method; (f) The signals by five-point sliding mean filtering method

Fig. 11. The logic diagram of the sliding correlation filtering algorithm

Fig. 12. Filter state transition diagram

Fig. 13. Data read/write and window sliding calculation

Fig. 14. Module timing simulation

Fig. 15. FBG demodulation system

Fig. 16. Original spectral signal and signal after sliding correlation filtering