中文摘要：

许多物体检测工作都涉及目标物体的面内旋转问题.为了提高多角度物体检测的精度,在Gualdi等提出的以物体位置和尺度的概率密度采样为核心的粒子窗口方法基础上,提出充分利用不同角度物体分类器的相关性,通过用一个角度的分类器来增强另一个角度分类器所对应的采样概率密度,使得概率密度更快的收敛到待检测物体的位置和尺度上.在人手检测上的实验表明,所提方法比传统基于遍历滑动窗口的方法和基于现有基于粒子窗口的方法有更高的检测效率和精度.

英文摘要：

In-plane rotation is a common problem in object detection.Gauldi et al.proposed a method named particle window,which used multi-stage sampling to update the probability density of the position and scale of objects.Based on that algorithm,this paper presents a method that uses the correlative information of different angle classifiers to enhance the precision of multi-angle object detection.This method uses the probability density of one classifier to boost others＇,which causes the probability density of object converges faster.The experimental results on hand detection show that the proposed method can effectively improve the precision and efficiency in contrast to traditional sliding window based and traditional particle window based methods.