Shot Boundaries Detection Based Video Summary Using Dynamic Time Warping and Mean Shift
Ahmed Khazaal Sulaiman, Sawsen Abdulhadi Mahmood · 2020
Recently, an increasing in videos amount presents in cyberspace in conjunction with growth of multimedia communications and technologies, and low-cost storage devices; it has become obligatory to have the capability to rapidly browse and search through huge multimedia databases. In this context, video summarization-based video retrieval has become a critical issue needs to be addressed. The main objective of shots boundary detection SBD methods pursues to detect the transitions between consecutive frames along video sequences in term of shots boundary detection. Hence, shots with abundant information are employed in the content-based video retrieval. This paper presents a new approach for shot boundary detection with two main challenges; static and dynamic objects with static and dynamic camera motion. The proposed method employed the Dynamic Time Warping DTW method for identifying the differences between sequenced frames. The abrupt and gradual changes are captured along frame's sequences within and between shots based on mean shift technique. From the experiments, the proposed shot boundary detection based key frame extraction method exhibits an efficient performance represented by more informative frames with less redundancy.