Abstract—In this paper, an efficient approach is proposed to
improve detection efficiency of sliding window based detection
methods by setting adaptive thresholds for regular object
detection in the moving environment. In the proposed approach,
the symmetry and variance (SYM-VAR) information of targets
is learned from current frame and historical frames, and the
information is used to filter out the sub-windows which may not
contain the targets in the next frame. Our experimental results
have demonstrated that the proposed approach can reduce
nearly 50% of the average detection time with a small tradeoff
of accuracy compared to typical HOG-based (histogram of
oriented gradient) methods.
Index Terms—Adaptive online learning, variance, symmetry,
sliding window detection.
The authors are with the Communication and Information Security Lab,
Shenzhen Graduate School, Peking University, China (e-mail:
zhuys@pkusz.edu.cn).
[PDF]
Cite:Zhenming Nong, Yuesheng Zhu, and Hao Lai, "A Rapid Pretreatment Method For Object Detection in
Dynamic Scenes," International Journal of Future Computer and Communication vol. 3, no. 2, pp. 119-123, 2014.
