Peak localization is a critical step in processing fiber Bragg grating (FBG) reflected signal. Aiming at the feature of FBG reflected signal, we propose a peak detecting algorithm applied to FBG reflected signal to solve the peak overlapping problem in peak localization. According to the ridge peak detecting algorithm based on continuous wavelet transform, the correspondence between wavelet coefficients and FBG reflected signals is analyzed, and the ridge peak detecting algorithm is improved to find the peak of overlapped FBG reflected signal. The experimental result shows that the improved wavelet ridge peak detecting algorithm can effectively solve the peak overlapping problem of FBG reflected signal. Through this algorithm, the error of peak detecting can be reduced, and the accuracy of peak localization can be improved.