In this chapter, a smart framework is presented for the object localization on a given ground-plane test map using heterogeneous stereo vision. In particular, the image pairs are captured by using static and Pan Tilt Zoom (PTZ)cameras, which are heterogeneous in terms of imaging parameters thus having different focal lengths, image resolutions, intensities, etc. These two cameras are selected in a co-operative manner from a network of static and PTZ cameras and used as a stereo system to localize an object even in the case when it is partially occluded. The various sequences of images captured by these cameras are made homogeneous based on their focal ratio and then by performing zero padding. The pairs of matching points are obtained using scale invariant features (SIFT) matching from stereo images. The rectification transformations are calculated by solving a constrained nonlinear optimization problem. The 3-D position of the object is estimated based on a modified concept of stereo matching in rectified pairs of images. Localization is made using the 3-D position obtained from stereo. Experiments are performed to evaluate the performance of the proposed framework using real sequences of images. The proposed method is useful in stereoscopic as well as in video surveillance applications.

Stereo vision in a network of co-operative cameras

MICHELONI, Christian;FORESTI, Gian Luca
2010-01-01

Abstract

In this chapter, a smart framework is presented for the object localization on a given ground-plane test map using heterogeneous stereo vision. In particular, the image pairs are captured by using static and Pan Tilt Zoom (PTZ)cameras, which are heterogeneous in terms of imaging parameters thus having different focal lengths, image resolutions, intensities, etc. These two cameras are selected in a co-operative manner from a network of static and PTZ cameras and used as a stereo system to localize an object even in the case when it is partially occluded. The various sequences of images captured by these cameras are made homogeneous based on their focal ratio and then by performing zero padding. The pairs of matching points are obtained using scale invariant features (SIFT) matching from stereo images. The rectification transformations are calculated by solving a constrained nonlinear optimization problem. The 3-D position of the object is estimated based on a modified concept of stereo matching in rectified pairs of images. Localization is made using the 3-D position obtained from stereo. Experiments are performed to evaluate the performance of the proposed framework using real sequences of images. The proposed method is useful in stereoscopic as well as in video surveillance applications.
2010
9781441909527
9781441909527
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11390/1095716
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact