The use of the mineral oil is an important cause of air pollution such as fog. The effectiveness and rapidity of the de-noising processing in mineral oil fluorescence spectroscopy detection system is a hot issue of the online real-time monitoring system. The de-noising method of the lifting wavelet transform (LWT) in the application of mineral oil fluorescence spectrum is proposed. Compared with traditional discrete wavelet transform (DWT), this wavelet transform method decomposes the existing wavelet filter module into the basic construction modules and steps to complete the transform with simplicity and a fast speed. There are characteristics of low computational complexity, in situ operation and the easy implement in the denoising process of mineral oil fluorescence spectra. The LWT can effectively solve the problems in these respects. The three methods of LWT, DWT and EMD are applied to the fluorescence spectra of 0# diesel oil, 97# gasoline and kerosene. The indicators evaluating de-noising effect such as the Signal-to-Noise Ratio (SNR), Mean Squared Error (MSE) and Normalied Correlation Coefficient (NCC) of the three kinds of mineral oil in the fluorescence spectra denoising prove the effectiveness of the lifting scheme wavelet transform in the application of mineral oil fluorescence spectrum. Meanwhile, the lifting scheme transform can improve the flexibility of structure and operation simplicity that makes the de-noising time reduced by 62%, validating the speediness of the de-noising method of the LWT in the application of mineral oil fluorescence spectrum and it is suitable for mineral oil fast de-noising processing system in real time.