计算机工程与应用 ›› 2017, Vol. 53 ›› Issue (19): 1-7.DOI: 10.3778/j.issn.1002-8331.1706-0322

• 热点与综述 • 上一篇    下一篇

方向梯度直方图综述

刘方园,王水花,张煜东   

  1. 南京师范大学 计算机科学与技术学院,南京 210023
  • 出版日期:2017-10-01 发布日期:2017-10-13

Survey on histogram-based optimization

LIU Fangyuan, WANG Shuihua, ZHANG Yudong   

  1. School of Computer Science and Technology, Nanjing Normal University, Nanjing 210023, China
  • Online:2017-10-01 Published:2017-10-13

摘要: 通过总结多篇基于HOG(方向梯度直广图)特征应用的文献,详细阐述了每篇文献所使用的方法,并对其实验结果进行了分析。HOG特征是一种经典特征描述方法,它通过不断改进,引入了PHOG、AW-HOG、MultiHOG对检测目标进行准确表示。通过分析HOG特征在人体行为检测、交通领域和新兴产业中取得的良好检测效果,说明HOG特征的发展前景值得期待。

关键词: HOG(方向梯度直广图)特征, 人体行为检测, 交通领域

Abstract: This paper provides a literature survey on the fundamental and application of Histogram of Oriented Gradients(HOG). Meanwhile, this survey elaborates the HOG and analyzes the experimental results of each literature. As is known, HOG feature is a traditional feature description method. The researchers continually improve HOG feature, to identify targets more accurately. The researchers have put forward PHOG, AW-HOG, MultiHOG and so on. The development prospect of HOG features is worth expecting for its remarkable detection results by applying it to different domains, such as human activity, transportation and other emerging fields.

Key words: HOG feature, human activity detection, transportation field