Data fusion considers discriminate performance made by different features
and gets a new discriminate vector which had higher classifiable power. To get better effect of combination
weighting combination was used to form the parts of with balance. Geometrical features of lip region and the descriptors of lip contours by Discrete Cosine Transform were combined to get a new discriminate vector. With this new vector
the HMM model was used to training and recognizing. The experimental results based on isolated Chinese words show that the new discriminate vector by weighting combination can produce better recognition result than using single characterize parameters when applied to lipreading. A maximum recognition rate was improved by 2.5% with different weighting factors.