2012 Fiscal Year Final Research Report

A study on construction of nighttime pedestrian image database and pedestrian detection method

Research Project

Project/Area Number	22500161
Research Category	Grant-in-Aid for Scientific Research (C)
Allocation Type	Single-year Grants
Section	一般
Research Field	Perception information processing/Intelligent robotics
Research Institution	Akita Prefectural University
Principal Investigator	SARUTA Kazuki 秋田県立大学, システム科学技術学部, 准教授 (80282193)
Co-Investigator(Kenkyū-buntansha)	CHEN Guoyue 秋田県立大学, システム科学技術学部, 教授 (20282014) TERATA Yuki 秋田県立大学, システム科学技術学部, 助教 (40360002)
Project Period (FY)	2010 – 2012
Keywords	コンピュータビジョン
Research Abstract	In this research issue we have developed a video database of pedestrians at night for research and development. In addition, we considered methods for pedestrian candidate region extraction and pedestrian detection in the regions. We shot the pedestrian image of more than 2,000 people and classified according to their content. We also confirmed the effectiveness and challenges of the candidate region extraction method using a moving amount and direction of movement of the pedestrian. Furthermore, we proposed the background rejection method and the pedestrian detection method that can select and visualize useful features to detect pedestrian, and showed the efficacy for nighttime pedestrian detection.

(3 results)