Raman signal, especially signal from living organisms, sometimes is interfered so seriously by the background fluorescence that it is too complex to get good result. According to the characteristics of fluorescence background, a new fluorescent background method based on min-max signal adaptive zooming is developed. To verify the feasibility of this method, two ways are used: one is the relation between peak intensity and concentration, the other is the relation between single spectrum and mixed spectra. Compared to the traditional fluorescent background elimination (polynomial fitting baseline correction), the results show that this method is highly feasible and superior to the traditional method. To image the 2D, 3D blood distribution in mice ear tissue in vivo, the peak at 1549 cm-1 which is one of characteristic peaks of hemoglobin is selected. By comparing the image results generated by this new method with the former process way, the results show this method can optimize the Raman spectral data, especially for low-density Raman characteristic peaks, so the imaging results are better.