A novel image rotating algorithm tailored for embedded AFIS

In recent years, the importance of information security in wide spread embedded applications increases a lot, promoting the study on fingerprint recognition, which is an effective biometric technique used in personal identification. In this paper, we study the effect of fingerprint image rotation on the correct recognition rate of automatic fingerprint identification system (AFIS), proving it a necessary consideration. In addition, a novel image-rotating algorithm is presented, which is efficient and compact, suitable for ASIC implementation. Based on this algorithm, we improve an existing AFIS, which originally can´t be applied in practice due to the unacceptable high misidentification rate. Experiment results show that the average recognition rate can be improved by 46.64% under the same condition, while the overhead is negligible.