GESTURE DESCRIPTION AND STRUCTURE OF A DICTIONARY FOR INTELLIGENT COMMUNICATION OF SIGN LANGUAGE IMAGES

In the intelligent transmission of the sign language image based on th e semantic information, one possible method is to transform the semant ic information on the receiving side into the motion parameters of the upper extremities and to synthesize the animation image of the sign l anguage. For this purpose, the word dictionary to describe the motion of the upper limb corresponding to the semantic information must be co nstructed. A problem then is that the human upper limb has a large deg ree of freedom. It is difficult to control skillfully the posture of t he upper limb using the motion parameters for the joint movements. Thu s, the complex motion is realized by combining the posture and motion of the upper limb, corresponding to the sign language. This paper prop oses a method in which the shape of the hand to be used in the sign la nguage is described by a table and the motion of the upper limb is des cribed and represented by the motion features of the gesture based on the joint angle model. The method to construct two kinds of word dicti onaries, called ''representative dictionary'' and ''realizing dictiona ry,'' are proposed to generate the animation image for the sign langua ge based on the semantic information. The representative dictionary gi ves the basic motion of the sign language. The realizing dictionary de scribes the control parameters such as the iteration and the speed of motion. The word in sign language is realized by adding the realizing form to the representative form. in the proposed method, the descripti on of the word in the signal language is simplified, which makes it ea sy to control the upper limb posture as well as to iterate or emphasiz e the motion. using the method proposed in this paper, a word dictiona ry is constructed to describe the gesture of the sign language. A syst em is discussed in detail which transforms the input sentence in Japan ese into the animation image.