A Novel Method on Summarization of Video Using Local Ternary Pattern and Local Phase Quantization

Abstract

In last decade, Video Summarization (VS) approach is playing a pivotal role in the analysis of the Video contents. The methodologies involved in Video Summarization have wide range of applications in the field of defense for video surveillance, intrusion, object detection, Video Browsing, Content-based Video Retrieval and Storage etc. In this study, we have proposed video summarization techniques to extract the frames of interest. Then, video summarization has determined by the advanced texture descriptors. Local Ternary Pattern (LTP) & Local Phase Quantization (LPQ) are the texture descriptor methods used to provide an efficient video summarization process. These methodologies are in conformity with the elimination of redundant frames in a video as well as the maintenance of user defined number of distinctive images. Then apply the clustering process, which is an unsupervised machine learning algorithms, such as, Affinity Propagation and BIRCH, are utilized to cluster the similar frames into one group. These methodologies confirm that the summary of video denotes the most distinctive frames of the input video, which results the same importance to preserve the continuousness of the summarized video.