Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Real-time moving pedestrian detection using contour features

Real-time moving pedestrian detection using contour features Pedestrian detection is one of the most fundamental research in computer vision. However, many high performance detectors run slowly. In this paper, we propose a real-time moving pedestrian detector by using efficient contour features. Firstly, the moving targets are detected by background subtraction. By combining the elliptic Fourier descriptors and the normalized central moments, we propose the Elliptic Fourier and Moments Descriptors (EFMD) to describe the moving target contours. Secondly, the moving targets are classified by the trained Support Vector Machine (SVM). In addition, we introduce a novel overlap handling algorithm based on linear fitting and normalized central moments, which improves the detection performance by reducing both false positives and miss rate. The experimental results on PETS 2009 and CAVIAR datasets show that our approach achieves a miss rate of 14% (PETS 2009) and 13% (CAVIAR) at 10−1 False Positives Per Image (FPPI) and an average runtime per frame of 30 ms (PETS 2009) and 25 ms (CAVIAR), which significantly outperforms several state-of-the-art detectors in both detection performance and runtime. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Multimedia Tools and Applications Springer Journals

Real-time moving pedestrian detection using contour features

Download PDF
20 pages

Loading next page...
 
/lp/springer_journal/real-time-moving-pedestrian-detection-using-contour-features-izMwYS6jUW
Publisher
Springer Journals
Copyright
Copyright © 2018 by Springer Science+Business Media, LLC, part of Springer Nature
Subject
Computer Science; Multimedia Information Systems; Computer Communication Networks; Data Structures, Cryptology and Information Theory; Special Purpose and Application-Based Systems
ISSN
1380-7501
eISSN
1573-7721
DOI
10.1007/s11042-018-6173-4
Publisher site
See Article on Publisher Site

Abstract

Pedestrian detection is one of the most fundamental research in computer vision. However, many high performance detectors run slowly. In this paper, we propose a real-time moving pedestrian detector by using efficient contour features. Firstly, the moving targets are detected by background subtraction. By combining the elliptic Fourier descriptors and the normalized central moments, we propose the Elliptic Fourier and Moments Descriptors (EFMD) to describe the moving target contours. Secondly, the moving targets are classified by the trained Support Vector Machine (SVM). In addition, we introduce a novel overlap handling algorithm based on linear fitting and normalized central moments, which improves the detection performance by reducing both false positives and miss rate. The experimental results on PETS 2009 and CAVIAR datasets show that our approach achieves a miss rate of 14% (PETS 2009) and 13% (CAVIAR) at 10−1 False Positives Per Image (FPPI) and an average runtime per frame of 30 ms (PETS 2009) and 25 ms (CAVIAR), which significantly outperforms several state-of-the-art detectors in both detection performance and runtime.

Journal

Multimedia Tools and Applications – Springer Journals

Published: Jun 2, 2018

References

  • Performance of Optical Flow Techniques
    Barron, JL; Fleet, DJ; Beauchemin, SS; Burkitt, TA
  • Semisupervised feature analysis by mining correlations among multiple tasks
    Chang, X; Yang, Y
  • Feature Interaction Augmented Sparse Learning for Fast Kinect Motion Detection
    Chang, X; Ma, Z; Lin, M; Yang, Y; Hauptmann, A
  • Bi-level semantic representation analysis for multimedia event detection
    Chang, X; Ma, Z; Yang, Y; Zeng, Z; Hauptmann, A
  • Semantic pooling for complex event analysis in untrimmed videos
    Chang, X; Yu, YL; Yang, Y; Xing, EP
  • Pedestrian detection: an evaluation of the state of the art
    Dollár, P; Wojek, C; Schiele, B; Perona, P
  • Fast feature pyramids for object detection
    Dollár, P; Appel, R; Belongie, S; Perona, P
  • A Bayesian Exemplar-based approach to hierarchical shape matching
    Gavrila, DM
  • Fourier preprocessing for hand print character recognition
    Granlund, GH
  • Deep belief networks
    Hinton, GE
  • Beyond trace ratio: weighted harmonic mean of trace ratios for multiclass discriminant analysis
    Li, Z; Nie, F; Chang, X; Yang, Y
  • Learning Multi-scale Block Local Binary Patterns for Face Recognition
    Liao, S; Zhu, X; Lei, Z; Zhang, L; Li, SZ
  • Contour-motion feature (CMF): A space–time approach for robust pedestrian detection
    Liu, Y; Chen, X; Yao, H; Cui, X; Liu, C; Gao, W
  • Distinctive image features from scale-invariant keypoints
    Lowe, DG
  • Example-Based Object Detection in Images by Components
    Mohan, A; Papageorgiou, C; Poggio, T
  • Pedestrian Detection with Spatially Pooled Features and Structured Ensemble Learning
    Paisitkriangkrai, S; Shen, C; Hengel, AVD
  • A trainable system for object detection
    Papageorgiou, C; Poggio, T
  • Faster R-CNN: towards real-time object detection with region proposal networks
    Ren, S; He, K; Girshick, R; Sun, J
  • A novel pixel neighborhood differential statistic feature for pedestrian and face detection
    Shen, J; Zuo, X; Li, J; Yang, W; Ling, H
  • Robust real-time face detection
    Viola, P; Jones, MJ
  • A blind watermarking scheme using new nontensor product wavelet filter banks
    You, X; Du, L; Cheung, YM; Chen, Q
  • Fast moving pedestrian detection based on motion segmentation and new motion features
    Zhang, S; Bauckhage, C; Klein, D; Cremers, A
  • An adaptive hybrid pattern for noise-robust texture analysis
    Zhu, Z; You, X; Chen, CL; Tao, D; Ou, W; Jiang, X; Zou, J