All protein blend materials have wide applications because they combine the excellent properties of the individual protein components. However, the phase behaviour, which is closely related to the properties of such blend materials, is difficult to be determined because of the similarity between the structures of each protein component. In this article, principal component analysis (PCA) and clustering analysis are introduced to analyse the FTIR spectroscopic imaging data of all protein blend materials. These attractive multivariate imaging approaches can provide more spectral information compared with the conventional univariate imaging (also called function group imaging) technique. Silk fibroin/soy protein isolation (SF/SPI) blend, a typical fibrous protein/globular protein blend, is selected as a sample material for investigation. The results clearly indicate that SF and SPI are phase separated in the blend, in which the globular SF domains are dispersed in the continuous SPI matrix. This is an interesting and surprising phenomenon because people always consider that the globular protein tends to form globular domains, whereas the fibrous protein easily forms a continuous matrix. Therefore, it is strongly suggested that the PCA and clustering imaging approaches have a remarkable ability to distinguish the phase behaviour in all protein blend materials, which may greatly help in improving the properties of existing materials and in the design of new materials.