Multilayer Video Semantic Feature Extraction Method
Xian Zhong, Tianbao Yu, Yansheng Lu · 2017
To solve video semantic retrieval, the key problem is the difficulty of analyzing, searching and processing the massive video data.In this paper, we propose an innovative multilayer video semantic feature extraction method.The whole method can achieve automatic extraction, label of the concept ontology of video semantic and provide support for the semantic retrieval.The main idea of the method includes the MPE (Model Parameter Estimation) algorithm, MR(Model Replace) algorithm and MVSS(Multilayer Video Semantic Symbolic representation) algorithm.Express the semantic information of the whole video key-frame from three levels.Furthermore, we examine the validity of those algorithms by labelling the video object ontology.The experimental results show that the method proposed in this paper can improve the accuracy rate of the video semantic feature extraction.