Title :
Flashlight Scene Detection for MPEG Videos
Author :
Wang, Jian ; Xu, Yanling ; Yu, Songyu ; Zhou, Yuanhua
Author_Institution :
Inst. of Image Commun. & Inf. Process., Shanghai Jiao Tong Univ.
fDate :
Oct. 30 2005-Nov. 2 2005
Abstract :
A new flashlight scene detection approach is presented. It focuses on two techniques: a new flashlight model based on the spatial-temporal characteristics of intensity value for the DC sequence, and a local threshold selection scheme using the sliding window. The advantages of this approach are its good performance for multi-frame and gradual flashlight scene cases, and local threshold selection. Experimental results show that the proposed algorithm is fast, robust and high accuracy
Keywords :
data compression; image sequences; spatiotemporal phenomena; video coding; DC sequence; MPEG video; flashlight scene detection approach; local threshold selection scheme; sliding window; spatial-temporal characteristics; Change detection algorithms; Content based retrieval; Gunshot detection systems; Image communication; Image edge detection; Image sequences; Information processing; Layout; Robustness; Videos; content-based video retrieval; flashlight scene detection; shot cut;
Conference_Titel :
Multimedia Signal Processing, 2005 IEEE 7th Workshop on
Conference_Location :
Shanghai
Print_ISBN :
0-7803-9288-4
Electronic_ISBN :
0-7803-9289-2
DOI :
10.1109/MMSP.2005.248676
