An Indexing Algorithm Based on Clustering of Minutia Cylinder Codes for Fast Latent Fingerprint Identification
Metadatos
Mostrar el registro completo del ítemEditorial
IEEE
Materia
Fingerprint indexing Latent fingerprint K-means clustering Minutia cylinder code
Fecha
2021-06-14Referencia bibliográfica
Pérez-Sánchez, I... [et al.] (2021). An indexing algorithm based on clustering of minutia cylinder codes for fast latent fingerprint identification. IEEE Access. [10.1109/ACCESS.2021.3088314]
Patrocinador
Consejo Nacional de Ciencia y Tecnologia (CONACyT) 492968Resumen
Latent ngerprint identi cation is one of the leading forensic activities to clarify criminal acts.
However, its computational cost hinders the rapid decision making in the identi cation of an individual when
large databases are involved. To reduce the search time used to generate the ngerprint candidates' order to be
compared, ngerprint indexing algorithms that reduce the search space while minimizing the increase in the
error rate (compared to the identi cation) are developed. In the present research, we propose an algorithm
for indexing latent ngerprints based on minutia cylinder codes (MCC). This type of minutiae descriptor
presents a xed structure, which brings advantages in terms of ef ciency. Besides, in recent studies, this
descriptor has shown an identi cation error rate, at the local level, lower than the other descriptors reported
in the literature. Our indexing proposal requires an initial step to construct the indices, in which it uses
k-meansCC clustering algorithm to create groups of similar minutia cylinder codes corresponding to the
impressions of a set of databases. K-meansCC allows for a better outcome over other clustering algorithms
because of the selection of the proper centroids. The buckets associated with each index are populated
with the background databases. Then, given a latent ngerprint, the algorithm extracts the minutia cylinder
codes associated with the clusters' indices with the lowest distance respect to each descriptor of this latent
ngerprint. Finally, it integrates the votes represented by the ngerprints obtained to select the candidate
impressions.We conduct a set of experiments in which our proposal outperforms current rival algorithms in
presence of different databases and descriptors. Also, the primary experiment reduces the search space by
four orders of magnitude when the background database contains more than one million impressions.