Wesam Ashour

Visual Speech Recognition: the Digits from One to Ten in the Arabic Language Wesam Ashour

A new algorithm to recognize the numbers from "one" to "ten" in the arabic language in this thesis. Analyzing a series of images for the speaker is the backboon of the thesis, detecting the first frame, the end frame, the motion vectore, and finally extracting the moment features from the motion vectore. A database consist of ten speakers are used in testing the results, every speaker pronounce the ten words fifteen times. Three recognition methods are used, K-mean, Fuzzy_kmean, and K-NN. The database is devided into two parts, the first part is to test which recognition method is the best for our purpose, while the second is for testing