Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Chapter
Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Chapter
}
TY - CHAP
T1 - Video object tracking with differential structural similarity index.
AU - Loza, Artur
AU - Wang, Fanglin
AU - Yang, Jie
AU - Mihaylova, Lyudmila
PY - 2011/5/24
Y1 - 2011/5/24
N2 - The Structural SIMilarity Measure (SSIM) combined with the sequential Monte Carlo approach has been shown [1] to achieve more reliable video object tracking performance, compared with similar methods based on colour and edge histograms and Bhattacharyya distance. However, the combined use of the structural similarity and a particle filter results in increased computational complexity of the algorithm. In this paper, a novel fast approach for video tracking based on the structural similarity measure is presented. The tracking algorithm proposed determines the state of the target (location, size) based on the gradient ascent procedure applied to the structural similarity surface of the video frame, thus avoiding computationally expensive sampling of the state space. The new method, while being computationally less expensive, has shown higher accuracy compared with the standard mean shift algorithm and the SSIM Particle Filter (SSIM-PF) [1] and its performance is illustrated over real video sequences.
AB - The Structural SIMilarity Measure (SSIM) combined with the sequential Monte Carlo approach has been shown [1] to achieve more reliable video object tracking performance, compared with similar methods based on colour and edge histograms and Bhattacharyya distance. However, the combined use of the structural similarity and a particle filter results in increased computational complexity of the algorithm. In this paper, a novel fast approach for video tracking based on the structural similarity measure is presented. The tracking algorithm proposed determines the state of the target (location, size) based on the gradient ascent procedure applied to the structural similarity surface of the video frame, thus avoiding computationally expensive sampling of the state space. The new method, while being computationally less expensive, has shown higher accuracy compared with the standard mean shift algorithm and the SSIM Particle Filter (SSIM-PF) [1] and its performance is illustrated over real video sequences.
KW - Tracking
KW - structural similarity
KW - gradient ascent
M3 - Chapter
SN - 978-1-4577-0537-3
SP - 1405
EP - 1408
BT - Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
PB - IEEE
CY - Prague
ER -