DOI:
https://doi.org/10.14483/2322939X.10991Publicado:
2015-12-08Número:
Vol. 12 Núm. 2 (2015)Sección:
Investigación y DesarrolloHistograma del gradiente con múltiples orientaciones (hog-mo) detección de personas
Histogram of the Gradient with Multiple Orientations (Hog-Mo) People Detection
Palabras clave:
clasificación de personas, vector de características, HOG-MO, múltiples orientaciones, SVM, SIT. (es).Palabras clave:
People classification, characteristic vector, HOG-MO, multi-orientation, SVM, ITS. (en).Descargas
Resumen (es)
En el campo de la visión por computador el problema de la clasificación de personas aún permanece como un desafío abierto de investigación. Por lo tanto, en este trabajo se realizan las siguientes aportaciones. Primero se introduce un nuevo método de extracción de características basado en el descriptor HOG (Histogram of Oriented Gradient) con múltiples orientaciones del gradiente, calculado sobre partes del cuerpo humano, denominado HOG-MO. Luego se construye un clasificador utilizando HOG-MO y SVM, se verifica su desempeño al compararlo con otras propuestas del estado del arte mediante las curvas ROC, logrando un adecuado equilibrio entre tiempo de cómputo y tasa de clasificación. En seguida se construye un sistema mono-cámara de detección de personas que trabaja en múltiples resoluciones, en el espectro visible, bajo condiciones variables de iluminación y de escala. Este sistema ha sido probado sobre una base de datos de personas en ambientes urbanos, en el espectro visible (BD-AU), creada para el desarrollo de aplicaciones en sistemas inteligentes de transporte (SIT) para la detección de peatones.
Resumen (en)
In the field of computer vision, the people classification problem has not yet been fully resolved and there are still many unknowns to solve. This paper presents a new version of the HOG descriptor, which is based on multi-orientation and human body parts, using the SVM classifier. The ROC curves show the discrimination power of each case analyzed, compared with the state of the art algorithms. Finally, a system for people detecting in urban environments is presented. This detector has been tested on a database created for applications in Intelligent Transport Systems (ITS).
Referencias
D. Gerónimo, “A global approach to vision-based pedestrian detection for advanced driver assistance systems”, Tesis PhD, Universidad de Barcelona, 2009
A. Sashua, A. Gdalyahu y Y. Hayun, “Pedestrian detection for driving assistance systems, single frame classification and system level performance”. In Proc IEEE Intelligent Vehicle Symposium. [En línea]. Disponible en http://www.cs.huji.ac.il/~shashua/papers/iv04-ped.pdf
T. Gandhi y M. Trivedi, “Pedestrian protection Systems: Issuess, survey and challenges”, IEEE Transactions on Intelligent Transport Systems, vol. 8, no. 9, sep. 2007.
APROSYS Project, “Strategies for enhanced pedestrian and cyclist friendly design”, septiembre, 2009.
S. Munder y D. Gavrila, “An experimental study on pedestrian classification”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 8, no. 11, pp. 1863-1868, 2006.
F. Xu y M. Gao, “Human detection and tracking based on HOG and Particle Filter”, IEEE 3rd International Congress on Image and Signal Processing (CISP2010), pp. 1503-1507, 2010.
X. Wang, M. Wang y W. Li, “Scene-Specific Pedestrian Detection for Static Video Surveillance”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 2, pp. 361-374, 2014.
K. Min, H. Son, Y. Choe y Y. G. Kim, “Real time Pedestrian Detection Based on a Hierarchical Two-Stage Support Vector Machine”, IEEE Conference on Industrial Electronics and Applications (ICIEA), pp. 114-119, junio 2013.
D. Brazey y C. Gout, “An algorithm for automatic people detection from depth map sequences”. European Workshop on Visual Information Processing (EUVIP), pp. 1-6, diciembre 2014.
D. Fernández, “Stereo vision based pedestrian detection system for assisted driving”, tesis PhD, Universidad de Alcalá, 2008.
P. Kelly, N.E. O'Connor y A.F. Smeaton, “A Framework for Evaluating Stereo-Based Pedestrian Detection Techniques”, IEEE Transacctions on Circuits and Systems for Video Technology, vol. 18, pp. 1163-1167, agosto 2008.
M. Enzweiler y D. Gavrila, “Monocular pedestrian detection: survey and experiments”, IEEE Transacctions on Pattern Analysis and Machine Intelligent, vol. 31, no.12, diciembre 2009.
0. Y. Tetik y B. Bolat, “Detection of pedestrians from still images”, IEEE Conference on Signal Processing and Comunication Aplications (SIU2011), pp. 670-673, junio 2011.
C. Keller, M. Enzweiler y D. Gavrila, “A New Benchmark for Stereo-Based Pedestrian Detection”, IEEE Intelligent Vehicles Symposium (IV2011), julio 2011.
C. Hilario, “Detección de peatones en el Espectro Visible e Infrarrojo para un Sistema Avanzado de Asistencia a la Conducción”, tesis Doctoral, Universidad Carlos III de Madrid, 2008.
D. Gerónimo, A. Sappa, D. Ponsa y A. Lopez, “2D-3D based on board pedestrian detection system. Computer Vision and Image Understanding”, Computer Vision and Image Understanding, vol. 114, pp. 583-595, 2010.
G. Xu, X. Gu, L. Liu y Z. Wu, “Real time pedestrian detection based on Edge factor and histogram of oriented gradient”, IEEE International Conference on Information and Automation (ICIA2011), julio 2011.
N. Dalal y B. Triggs, “Histograms of oriented gradients for human detection”, IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1, 2005.
N. Dalal, “Finding People in Images and Videos”, tesis PhD, Institut National Polytechnique de Grenoble, 2006.
Z. Li, Y. Sun, F. Liu y W. Shi, “An Effective and Robust Pedestrian Detecting Algorithm & Symposia”, IEEE International Conference on Intelligent Transportation Systems. Beijing, pp.1215, octubre 2008.
H. Cao, K. Yamaguchi, T. Naito y Y. Ninomiya, “Pedestrian recognition using second order HOG feature”, Proceeding of the 9th Asian Conference on Computer Vision (ACCV’09), vol. part II, pp. 628-634, Xian-China, 2010.
N. He, J. Cao y L. Song, “Scale Space Histogram of Oriented Gradients for Human Detection”, IEEE International Symposium on Information Science and Engineering (ISISE), Shanghai, China, pp. 167-170, 2008.
D. Fernández, I. Parra, A. Sotelo, P. Revenga, S. Alvarez y M. Gavilán, “3D candidate selection method for pedestrian detection on non-planar roads”, IEEE Proceedings of the Intelligent Vehicle Symposium, pp. 13-15, junio 2007.
M. Enzweiler, A. Eigenstetter, B. Schiele y D. Gavrila, “Multi-Cue Pedestrian Classification with Partial Occlusion Handling”, IEEE International Conference on Computer Vision and Pattern Recognition, San Francisco, 2010.
G. Lian, J. Lai y Y. Yuan, “Fast Pedestrian Detection using a Modified WLD Detector in Salient Region”, International Conference on System Science and Engineering, junio 2011.
N. Pal, “Object-Background Segmentation Using New Definitions of Entropy[C]”, IEEE Proc., pp. 284-295, 1989.
N. Cristianini y J. Shawe-Taylor, “An Introduction to Support Vector Machines and other ker
C. Chung y C. Lin, “LIBSVM: A library for support vector machines”, ACM Transactions on Intelligent Systems and Technology, vol. 2, issue 3, pp. 27:1-27:27, 2011.
O. Ludwig, D. Delgado, V. Goncalves y U. Nunes, “Trainable Classifier-Fusion Schemes: An Application To Pedestrian Detection”, 12th International IEEE Conference on Intelligent Transportation Systems, vol. 1. St. Louis, pp. 432-437, 2009.
K. Woods y K. Bowyer, “Generating ROC curves for artificial neural networks”, IEEE eTransactions on Medical Imaging, vol. 16, no. 3, junio 1997.
2.png)
