Using timing to detect horror shots in horror movies

Title: Using timing to detect horror shots in horror movies
Creator: Xu, Min; Jin, Jesse; Luo, Suhuai; Huang, Qingming
Relation: Progress in Pattern Recognition p. 225-231
Relation: Advances in Pattern Recognition
Relation: http://www.springer.com/computer/image+processing/book/978-1-84628-944-6
Publisher: Springer
Resource Type: book chapter
Date: 2007
Description: Affective content directly attracts an user's attention, appreciation and memory which also provides feasible entry for video highlights. Different from the existing movie emotion detection which makes use of traditional features, we introduce timing to detect horror shots in horror movies. From movie's point of view, timing is an important feature of movies and an important part of their power to affect viewers' feelings and emotions which talks about duration and duration relationship between frames. Shot-length and motion intensity are used to represent duration and duration relationship respectively. Decision tree is used for classification. Later, post-processing reduces both the classification error and human errors in subjective labeling on experimental data. Experimental data includes 4.5 hours video from 3 horror movies. Experimental results show that timing is effective and efficient to detect horror shots.
Subject: horror shots; horror movies; timing; viewers
Identifier: uon:5848
Identifier: http://hdl.handle.net/1959.13/44678
Identifier: ISBN:9781846289446

		Thumbnail	File	Description	Size	Format