We propose a novel method to separate overlapping absorption lines in gas detection based on fiber laser intracavity absorption spectroscopy. The method combines continuous wavelet transform, linear regression analysis and chaos particle swarm optimization together to retrieve the respective absorbance distributions of overlapping lines. The method’s principle is outlined and the rules for parameter selection are given. The performance of the method is demonstrated by both the simulated and experimental data. When the method is applied to the mixed gas concentration calibration of CO and CO2, high linearity is obtained with linear correlation coefficients of 0.9996 and 0.9997 respectively. The relative errors of CO and CO2 concentration retrieval are no more than 0.93% and 1.13% respectively.