基于改进Hough森林的对象检测方法

计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (26): 4-6.

基于改进Hough森林的对象检测方法

李子龙1，2，刘伟铭1

1.华南理工大学土木与交通学院，广州 510640
2.徐州工程学院信电工程学院，江苏徐州 221008

收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-09-11 发布日期:2011-09-11

Object detection method based on improved Hough forests

LI Zilong1，2，LIU Weiming1

1.College of Civil Engineering and Transportion，South China University of Technology，Guangzhou 510640，China
2.College of Information and Electrical Engineering，Xuzhou Institute of Technology，Xuzhou，Jiangsu 221008，China

Received:1900-01-01 Revised:1900-01-01 Online:2011-09-11 Published:2011-09-11

摘要/Abstract

摘要： 基于Hough森林的对象检测是隐式形状模型（ISM）的改进，它借助随机森林完成广义Hough变换。为了进一步提高其检测效果，充分利用训练图像中对象位置是已知的知识，改进了经典的偏移量不确定性度量方法，并优化随机森林的投票，使在Hough空间中真正对象的位置获得更多投票和更高的投票值。实验验证了该方法相比于经典的方法，具有更准确的对象检测效果。

关键词: 对象检测, 隐式形状模型, Hough森林

Abstract: Recent object detection techniques employ the generalized Hough transform using random forests，which is a improved Implicit Shape Model（ISM）.To improve its effectiveness of object detection，this paper improves the classical offset uncertainty measure and optimizes the votes of random forests，exploiting the knowledge of the object locations in the training images，which can maximize the responses and probabilistic votes at the true object locations in the Hough space.Experimental comparisons demonstrate that the proposed method outperforms the classical Hough forest technique.

Key words: object detection, Implicit Shape Model（ISM）, Hough forest

李子龙1，2，刘伟铭1. 基于改进Hough森林的对象检测方法[J]. 计算机工程与应用, 2011, 47(26): 4-6.

LI Zilong1，2，LIU Weiming1. Object detection method based on improved Hough forests[J]. Computer Engineering and Applications, 2011, 47(26): 4-6.