Shapes to Synchronize Camera Networks Richard Chang richard.chang@isir.fr Siohoi Ieng sio-hoi.ieng@upmc.fr Ryad Benosman ryad.benosman@upmc.fr Institut of Intelligent Systems and Robotics - CNRS, University Pierre and Marie Curie - Paris6 Abstract The synchronicity is a strong restriction that in some cases of wide applications can be difficult to obtain. This paper studies the methodology of using a non synchronized camera network. We consider the cases where the frequency of acquisition of each element of the network can be different. The following work in- troduces a new approach to retrieve the temporal syn- chronization from the multiple unsynchronized frames of a scene. The mathematical characterization of the 3D structure of scenes, is used as a tool to estimate syn- chronization value, combined with a statistical stratum. This paper presents experimental results on real data for each step of synchronization retrieval. 1 Introduction The synchronization operation is a task that com- plexifies many vision operations as the number of cam- eras becomes higher : cameras calibration, 3D recon- struction, frames synchronization, etc... Baker and Aloimonos [1], Han and Kanade [4] introduced pio- neering approaches of calibration and 3D reconstruc- tion from multiple views. Works on synchronization of cameras from images can be found in [12, 9]. Their aim is to retrieve synchronization in order to compute correctly 3D structures from a set of cameras. A solu- tion is to set hardware synchronization as in [6]. But this kind of method cannot be appliable because of spa- tial constraints. In these cases, a software based syn- chronization can be a way to solve this problem. Most of the former works exclude cases of heavy and/or non linear desynchronization as in [10, 11], or set special constraints on the scene or on the geometry of the cam- eras [8, 2]. In this paper, we introduce a new synchro- nization technique from 3D shapes. From all available frames which can be synchronized or not, 3D struc- tures are computed regardless they are correct or not. We will show that correct ones are generated only from synchronized frames. We then introduce a statistical ap- proach which will show that correct shapes reconstruc- tions (synchronized frames) occur more frequently than distorted ones (non synchronized frames). We will also explain the characterization of shape that allows the discrimination between correct and wrong reconstructions is possible. This paper is organized as follows. Section two describes the formal approach of our method. In section three, we will describe the syn- chronization algorithm. The section four presents ex- perimental results of the synchronization of a camera network. 2 Problem formalization 2.1 Shape criterion for synchronization. It is reasonable to assume that correct reconstruc- tions are possible if frames are synchronized and that unsynchronized frames lead likely to distorted results. We will prove in this section that this assumption is mathematically true : ”correct reconstructions” is equivalent to ”synchronized frames” if observed objects are rigid bodies. This can be done by examining simple planar motions. Let P 1 ,P 2 ,P 3 and P 4 be four collinear points viewed by C R and C L of centers O R and O L (see figure 1). Since the P i are collinear, we have the following rela- tions : P 1 P 2 = KP 1 P 4 and P 3 P 2 = M P 3 P 4 (1) K and M are constant scalars and we define L = ||P 1 P 4 ||. When the cameras C R and C L are synchronized, we 978-1-4244-2175-6/08/$25.00 ©2008 IEEE