Quality Analysis of Fingerprint Images Using Local Phase Quantization

The recognition performance of Automatic Fingerprint Identification System (AFIS) is immensely affected by the quality of the input fingerprint images. In a low-quality fingerprint image, various spurious minutiae points may be detected which may degrade the recognition performance of the AFIS system. Effective analysis of the low-quality fingerprint images prior to the fingerprint matching stage can aid in improving the recognition performance of the system. In this work, low quality fingerprint images are identified using a well known local textural descriptors called local phase quantization (LPQ). The local texture descriptors are gaining popularity due to their excellent performance and flexibility in analyzing the texture patterns. The experimental evaluations are carried out on low quality fingerprint images of publicly available FVC 2004 DB1 dataset. The achieved results show the high performance and robustness of the proposed method. As the proposed method outperforms the current state-of-the-art fingerprint classification methods, it can be utilized as a quality control unit during the fingerprint acquisition phase of the AFIS. The proposed method also has an advantage of computing only a single feature for fingerprint quality classification which makes it simple and fast approach.