Persian Vowel Recognition Using the Combination of Haar Wavelet and Neural Network

lips movement is an important parameter in speech recognition. The Lip image segmentation has a significant role in lips movement analysis. Major problems that any vowel recognition (especially Persian) method is faced are low chromatic in lip region, low contrast luminance, overlap between the lip and facial skin color, and similarity between lips movement in some vowels after detecting lips. In this paper, a new automatic and quick approach for the lip extraction based on using the Haar wavelet is proposed. The output of this proposed approach is used as a feature vector for a hybrid neural network. The proposed algorithm for lip image segmentation uses the color space CIE L*U*V* and CIE L*a*b* in order to improve the contrast between the lip and the other face regions. After this step, the lips are modeled and a feature vector with longitudinal and angular parameters is extracted from the proposed lips model.