MULTI-MODAL BASED VIDEO ABSTRACT GENERATION METHOD AND APPARATUS THEREOF

摘要

PPROBLEM TO BE SOLVED: To provide a method and apparatus in which the semantic level of abstracts is maintained and a processing speed of video abstract information generation can also be accelerated, by detecting an event candidate section based on audio information, then after detecting final event sections based on visual information only for the event candidate section, and generating video abstract information for the final event sections. PSOLUTION: The method includes the steps of detecting an event candidate section from video data, based on audio information; detecting shot change information from the detected event candidate section; detecting final event sections from the detected event candidate section, based on the detected shot change information and visual information; and generating video abstract information by combining the final event sections. PCOPYRIGHT: (C)2007,JPO&INPIT