Key frame extraction algorithm for video abstraction applications in underwater videos

摘要

It is extremely time consuming for researchers looking for particular events of interest to manually search in the video database. Therefore, there is enormous scope in research in the field of automatic extraction of key frames from underwater video sequences. Analysis of underwater video poses many challenges to existing techniques in computer vision including camera movement, turbidity, uneven background, non uniform lighting, moving objects in the underwater and inaccessibility of environment. The main aim of this paper is to (i) describe about the existing methods for key frame extraction, (ii) emphasize the key challenges in underwater environment, (iii) propose a new method to extract key frames which suits for underwater environment, (iv) perform the automatic detection and characterization of events of interest in underwater videos.