Laser-Induced Breakdown Spectroscopy (LIBS) has been widely used in more and more fields as a method for qualitative or quantitative analysisas a result of its unique advantages, such as no sample preparation, fast analysis speed, multi-element detection simultaneously, remote analysis, etc.. The theoretical basis of this technique is laser-induced plasma. Accurate measurement of spectral parameters of the plasma (such as spectral line intensity and plasma temperature) is a prerequisite for qualitative or quantitative analysis by LIBS. However, due to the inherent performance of the instrument in practical experiments, the distortion of the acquired spectral signal can be caused, which limits the precise measurement or calculation of the plasma spectral parameters. In order to overcome the impact of instrument performance, the influences of the inherent shortcomings of the Echelle spectrograph and the transmission fiber used in the experiment on the background noise and the absolute intensity of the spectral line were analyzed. Then, the spectral background noise was deducted by the peak clipping algorithms, and the absolute intensity of the spectral line was corrected by using the standard spectral data of the radiometric calibration sources. The spectral line intensity and plasma temperature were compared. It has been shown that the intensity correction has a great influence on the spectral signal with the wavelength less than 380nm forthe alloy steel sample. After the background subtraction and intensity correction, the plasma temperature was reduced from 13 401.75 to 8 980.72 K, and the fitting coefficient of Boltzmann plot for the plasma temperature was increased from 0.60 to 0.91. Therefore, it is necessary to take the spectral pretreatment before spectral data processing. It provides the basis to obtain the reliable spectral data for the qualitative or quantitative analysis of the material composition.