International Seminar of Resource, Environment, And Marine In The Global Challenge 2015 University of Wijaya Kusuma Surabaya, 29-30 September 2015 2 APPLICATION OF KINSHIP VERIFICATION BASED ON FACIAL GEOMETRY USING NEURAL NETWORK Shofiya Syidada Universitas Wijaya Kusuma Surabaya, Indonesia shofiya@uwks.ac.id Abstract: Face detection and recognition is one of the most important problems in computer vision. Facial images are important feature for human beings. From this facial images we can acquire lots of information about an individual person such as sex, age, and race. Furthermore, human face is one of physical characteristics that can be used as a biometric identity of a person besides fingerprint, retina and voices. One of key characteristics that can be identified from somebody facial images that whether there is kinship between parents and children. This research propose a framework for verification of relationship between parents and children, that can be done automatically. This verification performed by using feature vector on a pair of parents-children facial images. These features are extracted based on facial geometry. Facial geometry technique is to detect face image by using relatives positions and sizes of the important component of face. Then, feature selection is done to select appropriate feature to verify the kinship of parents-children from the feature vector on experiment data. Thereafter calculation of similarity are performed from this selected feature. The experiment data using 501 facial image pairs of parent and children. Based on the experiment, the best accuracy of kinship verification result is obtained 67% by using neural network. Keywords: Face Recognition, Feature Extraction, Kinship Verification, Neural Network, Similarity 1. Introduction Face detection and recognition, becoming an important research topic in the domain of computer vision and pattern recognition. Research on automatic face recognition has been developed. This research topic involves multidisciplinary. The latest computer applications involving this topic is face detection, identification, recognition and facial expression analysis (Rajpathaka, et.al., 2009). Some automatic face recognition application is face recognition from a variety of different poses, facial recognition on face image with accessories (like hats, sunglasses, slayer), age prediction based on face image and facial expression recognition. Face recognition automatically done by using specific features of the face and the distance between features. Specific features which are the main features of the face are eyebrows, eyes, nose and mouth. While the distance between facial features such as distance between the eyes, distance between nose to mouth, and width of the face (Li, et.al., 2011). Face image is important part of human identity. In addition, the face is one of the physical characteristics, which can be used as a biometric identity of a person, besides fingerprints, iris, retina, voice (Jafri, et.al., 2009). From the face image, a lot of information about a person can be obtained such as gender, age and ethnicity or race (Martello, et.al., 2010). Rathi (2012) declares that everyone has similar facial structures geometrically. And visually, everyone has minor differences in facial characteristics. However, one of the important characteristics that can be identified from the face image manually is whether there is a family relationship between parent and child. Research on the identification of kinship between one person and another has done Laurence (2006). Laurence says that the visual similarity of facial characteristics, is one sign of the presence of a genetic relationship, which is not influenced by age and gender. The similarity of facial characteristics genetically located on the upper face especially the eyes. Whereas, according to Siyu, et.al. (2012), genetically, facial resemblance lies in the eyes, nose, and mouth. In addition, the process of kinship verification using learning transfer that is applying a face image of parents at the same age as his son. The big challenge in the face image analysis is verifying kinship between two different people. Therefore, this study conducted automatic identification of a parent-child facial images to determine the presence or absence of kinship between the two people is based on the geometry of the face.