Title :
Motion object detection of video based on principal component analysis
Author :
Zhang, Chao-Yang ; Pan, Bao-chang ; Zheng, Sheng-lin ; Cao, Xiao-Jun
Author_Institution :
Fac. of Inf. Eng., Guangdong Univ. of Technol., Guangzhou
Abstract :
As the traditional method of background subtraction is interfered greatly by external factors, such as light, this paper proposes a method which utilizes three adjacent frames of image sequences to compute the principal components at first , then extracts the second principal component to remove the background of image. Through employing the operation of mathematical morphology and an improved background subtraction, it can remove the background of image efficiently, and implement quick detection. The result of experiment proves that, this method can improve efficiency and precision of detection, and has its practical value in the video monitoring.
Keywords :
image motion analysis; image sequences; mathematical morphology; object detection; principal component analysis; video signal processing; background subtraction; image sequence; mathematical morphology; motion object detection; principal component analysis; video monitoring; video processing; Computerized monitoring; Image motion analysis; Image sequences; Machine learning; Motion analysis; Object detection; Optical filters; Optical noise; Optical sensors; Principal component analysis; Background updating; Bottom-hat transformation; Principal component analysis; Top-hat transformation;
Conference_Titel :
Machine Learning and Cybernetics, 2008 International Conference on
Conference_Location :
Kunming
Print_ISBN :
978-1-4244-2095-7
Electronic_ISBN :
978-1-4244-2096-4
DOI :
10.1109/ICMLC.2008.4620911