Abstract
Tests of the intelligibility of two types of feature-extracted images for deaf communication are reported, at data rates around 10 kbit/s. A mean score of 34% (based on word count) was obtained for 4 sequences; however, individual scores on some sequences were 100% correct. The results point to the need for better form identification and two-way system evaluations.

This publication has 1 reference indexed in Scilit: