Techniques for key frame extraction: Shot segmentation and feature trajectory computation

Video is a structured media and segmentation of it into basic temporal units is normally the beginning in the processing channel of content based retrieval. This paper looks into the various possible techniques for extraction of key frames from the video stream. As there have been many challenges faced by other researches and projects to properly execute or create a system which can perform video frame extraction and face recognition, this paper will try to resolve them in a more efficient manner. The paper focuses on streamlining all the requirements of a user to view only those scenes which he desires i.e. scenes of a particular character and also the scenes of a particular emotion. The techniques considered herein are Shot Boundary Detection and NFL-technique.