The overlapped latent fingerprint separation by using Fourier Optics (FO)

Latent fingerprints are one of the most important pieces of evidence left at a crime scene and can be linked to all individuals involved. Each person’s fingerprints are unique and permanent, becoming an ideal biometric trait for the personal identification by extracting and comparing characteristic points (minutiae) of ridges. The overlapping fingerprint cases are frequently encountered in a crime scene and causing a difficult interpretation for an investigator. The problem has been challenging for forensic scientists over a decade. The method proposed in this study for the separation of overlapped latent fingerprints is based on the well-known spatial filtering method in the Fourier Optics (FO). Instead of tackling the problem by experiment, an alternative and simple means of image processing was proposed and conducted. The working principles start form converting spatial domain patterns (an image of overlapped fingerprints) to spatial frequency domain patterns or power spectrum, filtering out unwanted components (unwanted fingerprint) by appropriate spatial filters, and finally converting the modified pattern back to spatial domain patterns (an image of suspect fingerprint). As a result, the final image is improved from its original state. The periodic pattern of ridges is the key that allows FO to be used in the separation of the overlapped fingerprints. In this work, the procedures described are simply performed by an open source software: ImageJ. The FO-based image processing technique satisfactorily demonstrated its ability to recover an individual fingerprint from overlapping fingerprints.

On-chip optical spatial-domain integrator based on Fourier optics and metasurface

Analog optical computing has been an innovation and research interest in last several years, thanks to the ultra-high speed (potential for real-time processing), ultra-low power consumption and capability of parallel processing. Although great efforts have been made recently, no on-chip optical spatial-domain integrator has been experimentally demonstrated, to the best of our knowledge. Based on Fourier optics and metasurface, we design and fabricate an on-chip optical integrator using silicon-on-insulator (SOI) platform. The proposed integrator is able to integrate the electric field in spatial domain. As a proof-of-concept demonstration, a representative optical signal is well integrated to the desired distribution. Compared with theoretical expectation, the similarity coefficients of the simulated and experimental results are 83 and 78%, respectively. The proposed scheme has potential of performing more complex and ultra-high-speed computing for artificial intelligence.