Trace Recognition with a Feature Plane in the Learning System of Sign Language

Abstract

In this paper, a method of trace recognition in sign language learning system using virtual reality technology is suggested. Conventionally, if the 3 D trace data of sign language with individual difference are to be treated directly, long time is required in the learning process because of the large amount of data. To solve the problem, a method capable of mapping the 3 D traces of sign language into the 2 D traces is suggested based on the characteristics of the traces, and a convenient method on the trace recognition with individual difference is developed. The effectiveness of the proposed method is confirmed from the result of the trace recognition experiment using neural network. © 2000, The Japan Society of Mechanical Engineers. All rights reserved.