In order to analyze the mask material and establish the identification method of mask material evidence, 107 mask fiber samples were detected using Raman spectrometer for the study of the influence of black, pink and blue dyes on the Raman spectrum of mask fiber. Based on principal component analysis and correlation test, masks of different brands and batches of the same brand were distinguished, and the prediction of unknown brand mask was realized. The results show that under the condition of the same Raman characteristic peak, the mask with the strongest correlation with unknown brand can be quickly found by analyzing the difference of Raman intensity of mask fiber and the score chart of principal component. In the crime scene simulation experiment, the distance between unknown brand and similar brand is the closest in the principal component score graph, and the correlation coefficients of Pearson, Kendall rank and Spearman are the largest, which are 0.932, 0.799 and 0.942, respectively. Based on this, the mask with high similarity to unknown brand is found. This method reduces the scope of material evidence search, which can provide the basis for public security to find clues.