A Novel Approach for Static Video Content Summarization using Shot Segmentation and k-means Clustering

Due to the widespread availability of the Internet and video-capturing mobile devices, thousands of videos are uploaded and downloaded per second. The use of mobile cameras for capturing videos has increased a lot; it creates a new problem in handling these complicated video objects in terms of storage and processing. An efficient way in this situation is to shorten the original video into abstract summaries. These short summaries will give freedom to the user as per their interest. The proposed video summarization technique identifies only significant frames (i.e., keyframes) and produces static storyboards using shot segments and clusters—these extracted keyframes aid in creating video summaries for the abstraction and summarising of video summaries. Here, a static video content summarization is implemented as a problem of choosing significant keyframes and producing a visual static storyboard to describe the entire video content with less redundancy and good quality summary. This research proposes a novel clustering method to identify the set of static video frames, mainly keyframes and defines the novel way of creating static video storyboards. The experimentation is done using Open Video Project (OVP) videos. The performance of the video summarization is tested in terms of percentage accuracy. The set of keyframes obtained by the proposed system is validated using the OVP storyboards as ground truth; the results of the proposed shot and clustering-based keyframes extraction technique for static video content summarization are closer to the OVP ground truth and are improved than the Delaunay Triangulation video summaries.