Compressed domain texture based visual information retrieval method for I-frame coded pictures

In this paper we introduce a new compressed domain texture based visual information retrieval method. The proposed method is for the spatially predicted I-frames in the H.264 video coding standard. In I-Frame coding various prediction modes can be used for spatially predicting pixels of a block from upper or left side adjacent pixels. The selected prediction mode for a block indicates the way in which pixels of that block are related to their neighboring parts. Hence, we propose to use the histogram of the prediction modes as a texture descriptor of compressed I-frames. Since the method is based on independent I-Frame coded pictures, it can be used either for video analysis of H.264 coded videos or image retrieval of the I-frame based coded images such as advanced image coding. The simulation results indicate that this method has superior performance and lower computational load compared to an efficient realization of pixel domain texture retrieval method based on Gabor filter. Moreover, it is also robust to variations in image and coding parameters. Hence, this method is a powerful tool for visual information analysis in a wide range of applications.