PENGENALAN WAJAH DUAL VISION DENGAN METODE HALF-JOIN MENGGUNAKAN STEREO VISION CAMERA; DUAL VISION FACE RECOGNITION WITH HALF-JOIN METHOD USING STEREO VISION CAMERA

EDY WINARNO, Agus Harjoko

2015 | Tesis | FMIPA

The main problem in face recognition system based on half-face pattern is how to anticipate pose and illuminance variation for improving recognition rate and decreasing recognition time. To solve this problem, we can use two lenses on the stereo vision camera in face recognition system. Generally, a stereo vision camera is used in a face recognition system to reconstruct dots of face then to produce a three-dimensional (3D) face image. However, the process of face recognition using 3D images requires more recognition time than the use of a two-dimensional (2D) face images. Stereo vision camera has left and right lenses that are able to produce a 2D image of each lens. Stereo vision camera in face recognition has the capability to produce two of 2D face images from different angles. Both the angles of the face image will produce a detailed image of the face and better lighting levels on each of the left and right lenses. This research proposes a model of face recognition using the method of joining two face images from left and right lenses from a stereo vision camera namely half-join method. Half-join method is used in face image normalization processing. The proposed of half-join method consists of two face images joining model, symmetric half-join and asymmetric half-join. In symmetric half-join, RoI (region of interest) of face image will be produced by dividing a half of face image symmetrically. In asymmetric half-join, a RoI of face image from left and right lenses are produced based on the center axis of each eye from eye detection. The cropping face image from asymmetric half-join model has different wide of cropping depending on eyes coordinate location. Based on the experiments, the proposed system shows that the symmetric half-join and the asymmetric half-join method are able to produce in average of 93.25% and 97.5% of face recognition rate. The asymmetric half-join has a better accuracy than face recognition system using single vision method. This study shows that the stereo vision camera has the capability to produce two of 2D face images from each lens that is able to be elaborated with 2D face image recognition technique and able to result high recognition rate and short recognition time.

Kata Kunci : Face recognition; half-join; symmetric; asymmetric; stereo vision