FMX (EEPIS FACIAL EXPRESSION MECHANISM EXPERIMENT): PENGENALAN EKSPRESI WAJAH MENGGUNAKAN NEURAL NETWORK BACKPROPAGATION

Dwijotomo, Abdurahman and Sulistijono, Indra Adji (2010) FMX (EEPIS FACIAL EXPRESSION MECHANISM EXPERIMENT): PENGENALAN EKSPRESI WAJAH MENGGUNAKAN NEURAL NETWORK BACKPROPAGATION. EEPIS Final Project.

Preview

PDF - Published Version Download (1796Kb) | Preview

Abstract

In the near future, it is expected that the robot can interact with humans. Communication itself has many varieties. Not only from word to word, but body language also be the medium. One of them is using facial expressions. Facial expression in human communication is always used to show human emotions. Whether it is happy, sad, angry, shocked, disappointed, or even relaxed? This final project focused on how to make robots that only consist of head, so it could make a variety facial expression like human beings. This Face Humanoid Robot divided into several subsystems. There are image processing subsystem, hardware subsystem and subsystem of controllers. In image processing subsystem, webcam is used for image data acquisition processed by a computer. This process needs Microsoft Visual C compiler for programming that has been installed with the functions of the Open Source Computer Vision Library (OpenCV). Image processing subsystem is used for recognizing human facial expressions. With image processing, it can be seen the pattern of an object. Backpropagation Neural Network is useful to recognize the object pattern. Subsystem hardware is a Humanoid Robot Face. Subsystem controller is a single microcontroller ATMega128 and a camera that can capture images at a distance of 50 to 120 cm. The process of running the robot is as follows. Images captured by a camera webcam. From the images that have been processed with image processing by a computer, human facial expression is obtained. Data results are sent to the subsystem controller via serial communications. Microcontroller subsystem hardware then ordered to make that facial expression. Result of this final project is all of the subsystems can be integrated to make the robot that can respond the form of human expression. The method used is simple but looks quite capable of recognizing human facial expression. Keyword: OpenCV, Neural Network BackPropagation, Humanoid Robot

Actions (login required)