A Matching Method for Latent Images with Scale Variations

Latent fingerprint collected from the crime scene is one of the most frequently used evidence by the General Directorate of Security (EGM) and the Gendarmerie General Command (JGK) in solving committed crimes. To this end, fingerprints obtained from the crime scene are compared with the prints found in a certain database (obtained from fingerprint reader sensors) to identify the owner of the trace. It is observed that the performance of existing methods decreases significantly in cases where the spatial resolution and orientation of the fingerprints obtained from the sensor do not match that of the latent images obtained from the crime scene. With the proposed method in this study, it is ensured that the latent recognition process works successfully in cases where there is a scale difference. Proposed method has been tested on real world images obtained from EGM and JGK, and compared to the earlier proposed methods for latent recognition.