Face Recognition Methods by Using Low Resolution Devices Marian Spilka, Gregor Rozinaj Institute of Telecommunications, Slovak University of Technology, Tlkovicova 3, 812 19 Bratislava, Slovakia spilka.m@gmail.com, gregor.rozinaj@Stuba.sk Abstract - This paper focuses on the face recognition based on low-cost devices. It contains the modern forms of the face processing in 3D. The first part of the paper shows the comparison of the portable scanners. In the next section the face recognition process is split into three phases - capturing, processing and classifying. The options how to capture the face are listed and also the common processing methods for the face recognition are described. The classification methods are showed at the end of the section. In the last part of the paper we present our design of face recognition based on the face profile line, the face polygon and the color features of the face. Keywords - Face recognition; MuItimodal Interface, User Identification I. INTRODUCTION Face recognition is still very popular in science. The first attempt to process a face by computer was in the 90s [1]. Since then there has begun an idea to determine a user by a computer through the user's face. However, nowadays, new approaches have still been arising. The reason is that 3D scanners are cheaper and more available than before. One of the milestones in the 3D scanning was an introduction of the Kinect on November 2010 1 , which offers a low cost 3D capturing. This step has brought 3D devices from laboratories to people's living rooms. The face recognition is resolved and in use, but there are still many factors like a facial position, capturing conditions, a quality ofthe device, a mimicry, a facial motion and the others, which have an impact on the results. Thus the aim is to use the low cost devices which the wide public can afford to get the relevant results and to increase a ratio of a cost and a quality. n. COMPARISION OF SCANNING DEVICES Scanning devices are used to get an object from the reality to the virtual reality. Their purpose is to digitize a scanned object. The technologies are being improved every day and a trend of a device quality is rising up. Today, there are lots of scanning devices based on different technologies and used for various aims, so there are many approaches how to scan an object. Most of them use laser or optical sensors. The quality of scanning devices is related to their accuracy. The more accurate the device is, the higher quality ofthe result is. Today, I https/lnewsmicrosoft.com/20101 11I04/the-future-ot:entertainment-starts- today -as-kinect-for -xbox -360-leaps-and-lands-at-retailers- nationwide/#sm.0000qhc7731 Ovddx 1 sfyohS4mrmxo the most accurate devices use laser but these devices are the most expensive. In the case that quality of the scan is not the highest priority there is another approach - lR sensor. TABLE I. TABLE OF SCANNING DEVICES Devices Properties 3D scanner Sense 3D XYZ iSense 4 printin!!2 scanner 3 Resolution I,Smm 0,9mm@0,Sm 0,9mm@0,Sm Depth resolution I,Smm Imm@O,Sm Imm@O,Sm Distance lS-2Scm 0,2 - 1,6m 0,4 - 3,Sm Size of object 10 - 70 cm 0,2 - 2m O,2m- 3m Depth image size 640x480px 640x480px 640x480px Color image size 640x480px 1920xl080px 320x240px Released 20lS 2014 2014 Price $ 199 $393,14 $393,14 TABLE II. TABLE OF SCANNING DEVICES II Devices Properties Microsoft Kinect HandySCAN 2 5 Resolution O,7Smm@O,Sm Depth resolution I,Smm@O,Sm Distance 0,S-4,Sm Size of object ? Depth image size SI2x424px Color image size I 920x1080px Released 2014 Price $ 199 2 http/lus.xyzprinting.com/us_enlProduct/3D-Scanner 3 http://www.3dsystems.com/shop/sense/techspecs 4 http://www.3dsystems.com/shop/isense/techspecs 700™ 6 0.050 mm 0.030 mm 0,1- 4m ? -2,7Sm ? ? 2014 $S6900 5 https:lldeveloper.microsoft.com/en-us/windows/kinect/hardware 6 http://www.creafonn3d.com/en/metrology-solutions/products/portable- 3d-scanner/technical- specifications-handyscan- 3d 58 th International Symposium ELMAR-2016, 12-14 September 2016, Zadar, Croatia 185