摘要：

为解决复杂图像中的目标检测与定位问题，提出一种基于随机森林的目标检测与定位算法。采用SIFT局部特征构造随机森林分类器，以一个决策树中的全部叶子节点构成一个树型结构的判别式码本模型，从而获得更可靠的概率Hough投票，加快目标检测速度。实验结果证明，该算法效率较高，可用于复杂场景下的目标检测与定位。

关键词: 随机森林, 结构模型学习, SIFT局部特征, 判别式码本模型, 概率Hough投票, 目标遮挡

Abstract:

In order to solve the object detection and localization in the complicated image, this paper presents an algorithm for object detection and localization based on random forest. The Scale Invariant Feature Transform(SIFT) local features are used to construct a random forest classifier. A tree-structured discriminative codebook model is constructed by all leaf nodes of a decision tree. The discriminative codebook is used to estimate the object’s location via a probabilistic computation called probabilistic Hough vote. Experimental results show that the proposed algorithm has higher efficiency, and can provide a better detection results in a complicated environment.

Key words: random forest, structural model learning, Scale Invariant Feature Transform(SIFT) local feature, discriminative codebook model, probabilistic Hough vote, object occlusion

中图分类号: 

• TP391