Pedestrian Detection Using Haar-Like Features Based on Visual Memory

摘要：

We present a method for pedestrian detection that simulates the process of how visual memory work. This is a new way to get better performance of pedestrian detection. Visual memory mainly contains three steps: Feature learning, feature memorization and feature imagination. In this paper, we use Haar-like feature subtraction, Adaboost classifier and cascade architecture detector to simulate the three steps of visual memory, which makes the pedestrian detection get intelligence of biological vision. From the result of this method, we discuss the future work of the research about visual memory.

关键词： Pedestrian detection Haar-Like Visual Memory Adaboost Biological Vision

作者: Mingwei GUO Yuzhou ZHAO Zhiling WANG Zonghai CHEN

作者单位: Department of Automation, University of Science and Technology of China, Hefei, China

会议类型: 国内会议

会议名称: 2012年第14届中国系统仿真技术及其应用学术年会

会议地点: 三亚

会议语种:英文

页码: 252-255

在线出版日期: 2012-08-11（万方平台首次上网日期，不代表论文的发表时间）