Detektor Kebohongan Dengan Analisa Gerakkan Mata Dan Perubahan Diameter Pupil Berbasis Video Kamera Dan Image Processing Menggunakan Metode Haar Cascade Classifier Dan Neural Network (multilayer Perceptron)

Abstract

Abstrak Berbohong adalah sifat yang tidak terpuji, semua manusia didunia ini pasti pernah berbohong. Berbohong boleh dilakukan untuk kebaikan, namun banyak sekali orang – orang yang menggunakan kebohongan untuk menguntungkan dirinya sendiri. Sangat dibutuhkan sekali alat untuk mendeteksi kebohongan, namun harganya yang sangat mahal dan memiliki komponen yang banyak membuat masyarakat sulit memilikinya. Untuk menyelesaikan tugas akhir ini penulis membuat sistem untuk mendeteksi kebohongan berbasis video kamera dengan parameter yang berikan yaitu pergerakkan bola mata dan perubahan diameter pupil menggunakan metode haar cascade classifier. Teori psikologi menyimpulkan, seseorang yang berbohong akan cenderung melihat kearah kanan dan akan terjadi pembesaran pupil 4% sampai 7%. Dengan metode haar cascade classifier dan Neural network (multilayer perceptron) didapat hasil akurasi sistem sebesar 87%. Kata kunci : Lie detector, Haar cascade classifier, Neural network, Multilayer perceptron, Video kamera, Pupil mata, Eye tracking. Abstract Lying is a trait that is not commendable, all humans in this world must have lie. Lying can be done for good sake, but there are a lot of people who use lies in the wrong way, for example to slander others or to benefit themselves. The lie detector is urgently needed nowadays, but the price is expensive and there a lot of its components which make it difficult to own for the society and the lie detector only belongs to state security organization. Therefore, the affordable and easy components lie detector is needed, so that the society can understand the tool and use it wisely. To finish this final assignment, the Author make a system to detect someone’s lie based on camera video by anlysing the given parameters, which are eye moving (eye tracking) and the change of pupil diameter. With the method of Haar Cascade Classifier and Neural Network (Multilayer Perceptron). Psychological theories conclude that, someone who lies will have certain characteristics, especially in the part of eye, such as enlarged pupil diameter of eyes, the eyelid does not blink when it says lies, and the movement of eyeballs which always moving to indicate someone is thinking something. These parameters are to be tested and taken with video cameras that are integrated with the software to be analysed whether someone is lying or not. With the method of haar casecade classifier and neural network (multilayer perceptron) get the accuracy of the research system 87%. keywords:Lie Detector, Haar cascade classifier, Neural network, Multilayer Perceptron, Camera video, Eye pupils, Eye tracking.