Illumination invariant face recognition is a key problem for automatic face recognition system. The proposed method first applies two-degree ULBP to extract illumination invariant feature on multi-block face. Second, our method combines those histogram features into the final identification characteristics in the right order. SVM is then applying to feature classification to realize the face recognition. The experiment on YaleB database shows that, the combination of multi-degree and multi-block method can achieves a high face recognition rate.