Research of Video Retrieval based on Image and Audio Feature

Recently as computer technology and multimedia information have developed, text information as well as various types of image information can easy to obtain and be stored. In this paper, we proposed the efficient method of context based image retrieval to extract the face image and audio features in the video sequence. The eigenvectors for face can be obtained by the principal component analysis in the face image data. The feature parameter of audio information can be collected by classification of the voiced, unvoiced and silence of the voice in speech interval through the frame energy and zero-crossing rate. After the histogram feature vectors collect the cross function and similarity measurement, the database with the set of several face image and voice is able to create by applying fuzzy integral and assigning the weights. Our experiments and result shows that the method is more efficient and suitable result than the conventional image retrieval method. Keywords—Image Feature; Audion Feature; Video Retrieval