Key Frame Extraction Using Content Relative Thresholding Technique for Video Retrieval

Abstract

With the continuous growth in the space of information sharing over the Internet, the sharing of video information is also growing. The video information is highly appreciable for the online distance education, security, and distance healthcare, interactive communication over the Internet and also in multimedia news systems. The information captured in the video format demands summarization for effective processing and efficient storage. In order to summarize any video information, the best possible strategy is extracting key frames from videos. A number of research attempts are made in order to establish the most efficient key frame extraction framework. Nonetheless, most of the parallel research outcomes are affected by either high or low key frame extractions. Thus, the demand from the modern research is to build an optimal framework to extract key frames from motion videos. The major challenges are identified in this work and addressed in the finest way possible. This work demonstrates the framework for few different cases such as object in motion, camera in motion or both in case of highly colour contrast video sequences. The results of this framework demonstrate lowest time complexity and higher level of information preservation compared to the parallel research outcomes.