Fingerprint classification is an important and challenging stage in fingerprint recognition because of the complex search of large database. The purpose of this step is to narrow down the search space of fine matching. In the paper, the authors present a novel method of synthesis feature vector (called VTH) to effectively classify the fingerprint images (including even poor quality fingerprint images). Actually, this technique computes classification energy field based on combining orientation field energy (ridge structure feature) with orientation consistency value (describes how well the orientations over a neighborhood are consistent with the dominant orientation). Then, a square grid was placed on classification energy field (its center point locates at the reference point and it aligned based on reference orientation) to construct invariant feature vector VTH (for rotation and translation). Support Vector Machine-SVM will process vector VTH to classify fingerprint images. The experimental results on the FVC2004 database show the effectiveness and superiority of the proposed method.