Vgrajeni sistem za razpoznavanje šarenice pri uporabi nadzora dostopa v pametnem mestu

ZADNIK, DAMJAN (author), Žemva, Andrej (mentor)

Doktorska disertacija predstavlja delo na razvojno-raziskovalnem področju vgrajenih sistemov za razpoznavanje šarenice pri uporabi nadzora dostopa v pametnem mestu. Zaradi biometričnih značilnosti očesne šarenice uvrščamo sisteme za nadzor dostopa, ki temeljijo na digitalni obdelavi slike očesne šarenice, med najzanesljivejše in točne tovrstne sisteme. Razvit je bil prototip naprave za nadzor dostopa na osnovi razpoznavanja šarenice, na katerem je bil implementiran algoritem zajemanja ostre slike očesa. Naprava samodejno zazna in osvetli oko uporabnika ter zajame ostro sliko očesa, ko se le-ta z očesom postavi pred središče kamere na razdaljo približno 10 cm. Izdelek je v obliki tulca dimenzij le 40 mm × 40 mm × 40 mm, zato ga bo možno, po implementaciji še ostalih faz razpoznavanja šarenice, razmeroma preprosto vgraditi v elektronska vrata pametnih mest prihodnosti. Pri razvoju naprave je bil uporabljen V-model razvoja, katerega izhodišče so bile specifikacije zahtev, ki so bile pripravljene na osnovi podrobnega pregleda stanja na širšem znanstvenem področju razpoznavanja očesne šarenice in na ožjem znanstvenem področju implementacije sistema za razpoznavanje šarenice na vgrajenem sistemu. Izhodišče za specifikacije zahtev je bil tudi pregled stanja na trgu naprav za nadzor dostopa na osnovi razpoznavanja šarenice. Pred začetkom razvoja je bila narejena funkcionalna specifikacija sistema s pripadajočo analizo tveganja, ki ji je sledila specifikacija arhitekture sistema. V nadaljevanju je nato predstavljena razvita aparaturna oprema, ki ji sledi opis algoritma za zajemanje ostre slike očesa. Algoritem temelji na osvetlitvi očesa z vidno in infrardečo svetlobo ter digitalni obdelavi pik barvnega slikovnega tipala brez infrardečega filtra v realnem času. Pri razpoznavanju šarenice je zmogljivost sistema zelo odvisna od kakovosti zajete slike očesa. Zato smo opravili raziskavo, katere cilj je bil določiti zmogljivost razvite naprave v fazi zajemanja ostre slike očesa. Zajeli smo 100 slik očes 50-ih prostovoljcev in na osnovi predlaganih meril za oceno kakovosti zajetih slik določili zmogljivost podsistema za zajemanje ostre slike očesa. Povprečen čas zajemanja slike je znašal 1,65 s, pri čemer je 61 % zajetih vzorcev ustrezalo predlaganim merilom ustrezne kakovosti zajete slike očesa (FTE = 39 %).

• Language: Slovenian
• Keywords: nadzor dostopa, razpoznavanje šarenice, merila za oceno kakovosti
• Work type: Doctoral dissertation
• Typology: 2.09 - Master's Thesis
• Organization: FE - Faculty of Electrical Engineering
• Year: 2024
• PID: 20.500.12556/RUL-165372
• COBISS.SI-ID: 219488259
• Publication date in RUL: 04.12.2024

Secondary language

• Language: English
• Title: Embedded system for iris recognition in smart-city access control applications

Abstract

This doctoral dissertation presents the research and development process in the field of access control systems based on iris recognition. Due to several positive biometric properties of the human eye, the iris recognition technology offers one of the highest security levels among similar access control modalities. A prototype of an access control device based on iris recognition was developed, featuring an algorithm for sharp eye image acquisition. The device automatically detects and illuminates the user’s eye when it is positioned at a distance of 10 cm from the camera and captures an image. The cylinder-shaped device, measuring only 40 mm × 40 mm × 40 mm can easily be integrated into any electronic door in a smart city, once all iris recognition phases are implemented. During the development of the device, the V-model design approach was followed. This design is based on the requirement specification document. The inputs for this document were gathered from a market competition overview and existing scientific state-of-the-art developments in the field of iris recognition and its implementations on various embedded systems. Before development began, the functional specification of the device was specified. This formed the basis for the risk management process and the specification of the device’s architecture. Next, the developed hardware equipment and a description of the implemented algorithm for sharp eye image acquisition are presented. The algorithm is based on simultaneous visible and infrared illumination and real-time digital signal processing of a color image sensor with no infrared filter. The performance of any iris recognition system depends significantly on the quality of the captured eye image. Therefore, research to assess the performance of the developed image acquisition subsystem was conducted. One hundred images of eyes from fifty volunteers were acquired using the developed device. The measured average acquisition time was 1.65 seconds. The Failure to Enroll rate of the device, based on the proposed iris image quality metric, was 39 %.

• Keywords: access control, iris recognition, iris image quality metric