Murvet Kirci

On ear biometrics

Today the most successful biometric based identification technologies such as fingerprint, iris, retina, palm and face recognition are used worldwide in both criminal investigations and high security facilities. These technologies are well-studied, but research shows they have many drawbacks which decrease the success of the methods applied. Ear images are not affected by emotional expression, illumination, aging, poses and alike. In this study principal component analysis (PCA), fisher linear discriminant analysis (FLDA), discriminative common vector analysis (DCVA), and locality preserving projections (LPP) were applied to ear images for personal identification. The error and hit rates of four algorithms were calculated by random subsampling and k-fold cross validation.

Effects of the facial and racial features on gender classification

This paper presents an experimental study on examining the effects of facial and racial features on gender classification. In order to show which facial feature is the most influential for gender classification, parts of several face images, such as, forehead, eyebrows, eyes, nose, lip and chin were masked. For dimension reduction, Principal Component Analysis (PCA) and for determination of gender, Fisher Linear Discriminant (FLD) algorithms were applied to masked face images. Moreover, the effects of racial features on gender classification were studied. Experimental results indicated that the nose is the most influential part for gender classification. Furthermore the gender of the Asian people is more easily distinguished than that of the people of African origin.

Verification of Session Initiation Protocol Using Timed Colored Petri Net

In this work, Session Initiation Protocol model is established by using Timed Colored Petri Nets (TCPN). SIP (Session Initiation Protocol) is a protocol developed to assist in providing advanced telephony services across the Internet. The Session Initiation Protocol (SIP) has become the quasi-standard for Voiceover-Internet Protocol (VoIP) communications. SIP is based on a client–server infrastructure in which user agents represent the end-terminals as clients, proxy servers handle SIP message routing between the user agents, and registrar servers store the client’s contact information into a location service. By use of timed color set and useful time attributes in tokens defined in CPN tools, timer and time-related problems of SIP are modeled and analyzed. Timer is an important part for SIP, especially the INVITE transaction.

ARM based ear recognition embedded system

This paper presents a fully embedded ear recognition system using Principal Component Analysis (PCA) based on an ARM microcontroller which can be programmed with MicroC language. As its a common belief, PCA is the first known and easiest way for pattern recognition, but its not used in many applications with a microcontroller unit until now. The ear recognition system is programmed and can be used independent from a regular computer as a fully embedded system. Dataset is stored in the flash memory and the system can be trained in every validation and test processes without a regular computer. Experimental results for recognition rates are acceptable and as good as for safety systems.

Design of a fully-differential double folded cascode class AB opamp with continuous time common mode feedback network for 12-bit pipeline ADC applications

This paper deals with a fully differential (FD) operational amplifier (opamp) consists of double folded cascode and class AB output with continuous time common mode feedback (CMFB) network. The opamp is designed in 0.18 μm CMOS technology in Cadence Spectre Circuit Simulator with 1.8 V supply. In this single supply, opamp has 117 dB gain, 65-degree phase margin, 72 dB common-mode rejection ratio (CMRR) for 2pF load with a power consumption of only 1.2 mW. This topology is more suitable for high-speed pipeline Analog-to-Digital converters (ADCs) with foreground calibration, when compared to conventional designs, enabling operation at higher clock frequencies as the class AB stage causes the slew limiting in the first stage and power dissipation is decreasing.

A Novel approach for automatic ship type classification

This work classifies the ship types from color images by using cameras mounted on ships. Our data set contains 10 different ship types. The synthetic images used for training imported from Google 3D Warehouse. Test data set imported from Google Images and contains real ship images. This work aims to classify real ship images by using synthetic images. We present a novel approach for combining four features extracted from synthetic images and we have achieved % 90 accuracy.

A comparison of PCA, LDA and DCVA in ear biometrics classification using SVM

Despite increasing three dimensional recognition rate in ear biometric, there is need for special equipment to three dimensional image. Ear biometrics recognition rate was obtained high success by combined distinctive common vector approach methods with support vector machines in two-dimensional low-resolution cameras used surveillance and security system. In particular, this method will provide an important contribution to the non-cooperative personnel identification.

Analysis of automated systems by using coloured Petri Nets

In this work, the analysis of behavioural and structural features of automated systems has been investigated by using coloured Petri Nets (CPNs). CPN is one of the best modelling methods of the automated system. Many features of Petri Nets can be searched by reachability tree and invariants. They also show the features of CPN. Reachability tree shows the markings which can be reached by the net. With P invariants we can determine the correctness of the net. In this work, an algorithm to calculate reachability tree of automated systems is developed. P invariants of CPN is analysed and Martinez-Silva algorithm is applied to CPNs. Time performances of these methods are also determined.