The work aims to propose a dot feature extraction method based on the bi-dimensional empirical mode decomposition (BEMD), in order to achieve the intelligent diagnosis of the abnormal dot state in the printing process. Through the BEMD analysis on the dot image, its 2D intrinsic mode component was obtained and its feature extraction was done with the gray-level co-occurrence matrix (GLCM), so as to construct the feature representation of the printing dot. A classification experiment was carried out through the decision-making method of SVM. The proposed method could accurately diagnose such abnormal states as improper dot pressure and uneven ink. The accuracy rate of the dot classification experiment reached over 90%. BIMF-GLCM analysis has a good representational capacity for dot features and the related research provides an effective method for the construction of intelligent diagnosis feature set in the printing dots.