Structured Local Edge Pattern Moment for pedestrian detection

Author

Su, Song-Zhi ; Chen, Shu-Yuan ; Li, Shao-Zi ; Duh, Der-Jyh

Author_Institution

Cognitive Sci. Dept., Xiamen Univ., Xiamen, China

fYear

2010

fDate

9-11 April 2010

Firstpage

556

Lastpage

560

Abstract

Local feature based approaches have gotten great success in object detection and recognition in recent years. In this paper, a novel local based feature, Structured Local Edge Pattern Moment (SLEPm), is proposed for pedestrian detection in the sliding window framework. SLEPm encodes not only the statistical information but also the structure and spatial information of object for pedestrian detection. Linear Support Vector Machine (SVM) is used as a binary classifier to determine whether a sub-window contains pedestrian. Experimental results in INRIA pedestrian database show that performance of SLEPm is better than that of Histogram of Oriented Gradient (HOG).

Keywords

feature extraction; object detection; pattern classification; support vector machines; INRIA pedestrian database; binary classifier; histogram of oriented gradient; linear support vector machine; local feature based approaches; object detection; pedestrian detection; sliding window framework; spatial information; statistical information; structure information; structured local edge pattern moment; Computer science; Face detection; Filters; Head; Image edge detection; Intelligent robots; Leg; Object detection; Support vector machine classification; Support vector machines; Linear SVM; Local Edge Pattern; Object Detection; Pedestrian Detection; Structured Local Edge Pattern Moment;

fLanguage

English

Publisher

ieee

Conference_Titel

Image Analysis and Signal Processing (IASP), 2010 International Conference on

Conference_Location

Zhejiang

Print_ISBN

978-1-4244-5554-6

Electronic_ISBN

978-1-4244-5556-0

Type

conf

DOI

10.1109/IASP.2010.5476054

Filename

5476054