Accuracy Amelioration of an Integrated Real-Time 3D Image Sensor

Jad Ayoub (1) , Olivier Romain (1) , Bertrand Granado (2) , Yasser Mohanna (3)

1 SYEL - Systèmes Electroniques
2 ETIS - UMR 8051 - Equipes Traitement de l'Information et Systèmes
3 LPM - Laboratoire des Physiques de Matériaux

In this paper we investigate an active vision technique implemented in an embedded system for 3D shapes reconstruction. The main objective of the work is to have a balance in the accuracy of all components in the system where the size and autonomy of such an embedded sensor are hard constraints. This is achieved through the improvement of the pre-processing algorithms by reducing the time needed to compute the spots centers. In addition, lens distortion of the camera is included in the model to increase accuracy when reconstructing objects. Experimental evaluation shows that the size and the time are reduced, precision increased, when the resources spent on processing are relatively acceptable in comparison to the benefits.

Mots clés: 3D reconstruction; Active stereovision; Camera calibration; Lens distortion

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]; Electronique

Jad Ayoub, Olivier Romain, Bertrand Granado, Yasser Mohanna. Accuracy Amelioration of an Integrated Real-Time 3D Image Sensor. Conference on Design & Architectures for Signal and Image Processing, Nov 2008, bruxelles, Belgium. ⟨hal-00584256⟩