Crease Detection and Repair Based on Minutia Density Distribution

The existence of crease fatally destroys fingerprint structure and texture, concretely, creases will dissever ridgelines and produce a series of pseudo minutiae. Thus, creases will reduce the accuracy of the fingerprint recognition algorithm obviously, especially in the leading algorithm based on minutiae-matching. In this work, we propose a novel approach for detecting and repairing the creases. First, from the minutia density distribution, we sift out large minutia density areas as the crease candidates. We select the candidates meeting some constraints as true crease areas, then we divide them into Large Orientation Difference Crease Areas (LODCAs) and Small Orientation Difference Crease Areas (SODCAs). Secondly, we reconnect the broken ridgelines using stepwise approximation approach in SODCA, while in LODCA, triangular constraint approach is more applicable. The experimental results indicate that the algorithm can detect and repair creases effectively and the recognition accuracy improves greatly with wee additional calculation.