Video object plane extraction for surveillance applications

Author

Li, Juan ; Zhang, Xian-Min

Author_Institution

Inst. of Image Process. & Pattern Recognition, Shanghai Jiao Tong Univ., China

Volume

6

fYear

2004

fDate

26-29 Aug. 2004

Firstpage

3928

Abstract

Extraction of video object plane (VOP) is the central challenge for content-based video processing. It enables separate and flexible manipulation, blending and arrangement of content at MPEG-4 based decoder. We present a fast and efficient extraction algorithm depending on edge detection. Our algorithm begins with finding edge candidates detected by Canny edge detector. After selecting edge pixels which belong to the moving object in current frame, the VOP is extracted by using morphological operations. The proposed algorithm is evaluated on MPEG-4 test sequence and generates desirable VOPs at a lower computation cost.

Keywords

edge detection; image resolution; surveillance; video signal processing; Canny edge detector; MPEG-4 based decoder; MPEG-4 test sequence; content-based video processing; edge pixel; extraction algorithm; flexible manipulation; surveillance application; video object plane extraction; Background noise; Data mining; Image edge detection; Image processing; Intelligent systems; MPEG 4 Standard; Morphological operations; Noise generators; Pattern recognition; Video surveillance;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on

Print_ISBN

0-7803-8403-2

Type

conf

DOI

10.1109/ICMLC.2004.1380536

Filename

1380536