RNN Based Bitstream Feature Extraction Method for Codec Classification

摘要

In this paper, we propose codec classification algorithm based on recurrent neural network (RNN) model. In videocompression, codecs, such as MPEG2 and H.264/AVC, have their own distinctive data structure. These unique structureswhich are almost shown in header can be considered their feature. The proposed algorithm exploits that characteristics forclassifying unknown bitstreams into specific codec. According to the fact that RNN is appropriate to time series data forlearning to classification/recognition, the feature of an encoded bitstream can be extracted. We constitute the encodedbitstream as an input and give the bitstream its label indicating codec index. Two standard codecs, MPEG2 and H.264/AVC,are used in experiment. Experimental results show that the proposed RNN model classified bitstreams into correspondingcodecs to some extent.