A Method of Summarizing News Video Based on Multimodal Features

摘要

PURPOSE: A method for summarizing a news video based on a multimodal feature is provided to divide the news video provided together with a closed caption by the news article, and to summarize the news video by the divided news article, to utilize the summarized news video in contents-based video browsing and utilize an index result of generated closed caption data in establishing a database for text-based video search. CONSTITUTION: A digital contents acquiring step(S110) converts broadcasting signals of a news video into digital signals, and stores the converted digital signals. A multimodal feature extracting step(S120) extracts features from each multimodal component configuring the acquired digital contents. A major section detecting step(S130) detects major sections of the news video based on the extracted multimodal features. A video summary describing step(S140) structurally describes summary information of the news video based on the detected major sections. In the digital contents acquiring step, the broadcasting signals of the news video are divided into audio visual signals and closed caption signals, if the news video is an analog type. Then, the divided audio visual signals are converted into the digital signals and closed caption signals are decoded and stored in a text type.