Multi-object tracking using Kalman filter and particle filter

• Published in: 2017 IEEE International Conference on Power, Control, Signals and Instrumentation Engineering (ICPCSI) pp. 1688 - 1692
• Main Authors: Bukey, Chetan M., Kulkarni, Shailesh V., Chavan, Rohini A.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2017
• Subjects: Estimation; Kalman filter; Kalman filters; Multiple Object tracking; Object detection; Object tracking; Optical filters; Particle filter; Particle filters; Target tracking
• ISBN: 9781538608135; 1538608138
• DOI: 10.1109/ICPCSI.2017.8392001

Tracking Object is essential step for image and video processing research area and in computer vision technology applications like object identification, traffic control, automated surveillance systems and navigation systems. Foreground image separated from background image by conventionally image processing techniques. Background subtractions utilizing Gaussian Mixture Model (GMM) is basically utilized as a part of extricating elements of moving items and takes information in frames. The outcome demonstrates that GMM performs well when obstructions are there. Multiple objects tracking have been done using two methods that is Kalman filter (KF) tracking and the Particle filter (PF) tracking. The KF evaluate present, previous, and even future condition of object. Also Kalman filter can estimate even when exact idea of the demonstrated framework is unknown. PF have been being exceptionally helpful in multiple objects tracking for non-Gaussian and nonlinear estimation problems. The algorithm applied effectively on standard video database of PETS.