#TITLE_ALTERNATIVE#

• Main Author: Tanjung sekar (NIM : 232 99 110), Evita
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/7849
• Institution: Institut Teknologi Bandung

Abstract: <br /> <br /> <br /> <br /> <br /> Deaf people are hearing by seeing lipmovement, and then they need a special training to learn to speak, which as we know this special training is not satisfied enough. When they communicate among themselves, they use sign language. Sign language can be developed also in the communication between deaf people and the normal people with a special tool. <br /> <br /> <br /> <br /> <br /> The other research on developing these special tools, have already done in some countries, but there is no publicated research on developing these special tools to translate Indonesian sign language to text until this research was done. In the research, we tried to design a special tool to detect position of right hand, which is indicating words in sign language. The tool consist of series of flexure censor integrated on the glove, which will detect information of right hand position in analog voltage. The information will be converted into digital information by ADC and then will be processed by patern recognition. We use artificial neural network with reduced memory Levenberg-Marquardt which is variation of back propagation as learning method in the process of pattern recognition. The output of this process are word of sign language in text format. <br /> <br /> <br /> <br /> <br /> We use two cases to test our system performance. In the first case, system was successfrilly recognize 83,18% pattern of words which were produced by differencies between the position of wrist, thumb, index finger, middle finger, ring finger and little finger. But in the second case, system only recognize 49,58% pattern of words words which were produced by differencies between the position of shoulder joint, elbow, wrist, thumb, index finger, middle finger, ring finger and little finger.