Video Editing Based on Situation Awareness from Voice Information and Face Emotion

Video camera systems are becoming popular in home environments, and they are often used in our daily lives to record family growth, small home parties, and so on. In home environments, the video contents, however, are greatly subjected to restrictions due to the fact that there is no production staff, such as a cameraman, editor, switcher, and so on, as with broadcasting or television stations. When we watch a broadcast or television video, the camera work helps us to not lose interest in or to understand its contents easily owing to the panning and zooming of the camera work. This means that the camera work is strongly associated with the events on video, and the most appropriate camera work is chosen according to the events. Through the camera work in combination with event recognition, more interesting and intelligible video content can be produced (Ariki et al., 2006). Audio has a key index in the digital videos that can provide useful information for video retrieval. In (Sundaram et al, 2000), audio features are used for video scene segmentation, in (Aizawa, 2005) (Amin et al, 2004), they are used for video retrieval, and in (Asano et al, 2006), multiple microphones are used for detection and separation of audio in meeting recordings. In (Rui et al, 2004), they describe an automation system to capture and broadcast lectures to online audience, where a two-channel microphone is used for locating talking audience members in a lecture room. Also, there are many approaches possible for the content production system, such as generating highlights, summaries, and so on (Ozeke et al, 2005) (Hua et al, 2004) (Adams et al, 2005) (Wu, 2004) for home video content. Also, there are some studies that focused on a facial direction and facial expression for a viewer’s behavior analysis. (Yamamoto, et al, 2006) proposed a system for automatically estimating the time intervals during which TV viewers have a positive interest in what they are watching based on temporal patterns in facial changes using the Hidden Markov Model. In this chapter, we are studying about home video editing based on audio and face emotion. In home environments, since it may be difficult for one person to record video continuously (especially for small home parties: just two persons), it will require the video content to be automatically recorded without a cameraman. However, it may result in a large volume of video content. Therefore, this will require digital camera work which uses virtual panning and zooming by clipping frames from hi-resolution images and controlling the frame size and position (Ariki et al, 2006). Source: Digital Video, Book edited by: Floriano De Rango, ISBN 978-953-7619-70-1, pp. 500, February 2010, INTECH, Croatia, downloaded from SCIYO.COM