In this paper we present a system for 3D reconstruction of traffic scenes. Traffic surveillance is a challenging scenario for 3D reconstruction in cases, where only a small number of views is available that do not contain much overlap. We address the possibilities and restrictions for modeling such scenarios with only a few cameras and introduce a compositor that allows rendering of the semi automatically generated 3D scenes. Some of the occurring problems concern camera images, which might show a common background area, but can still differ drastically in lighting effects. For foreground objects nearly no common visual information might be available, as angles between cameras may exceed even 90°

No abstract

No abstract

This paper discusses new methods for coding of multi-view images. Three distributed solutions are proposed based on a parent node and child node framework. A parent node encodes the whole image whereas a child node only partially. The proposed scheme allows independent encoding of each view. Experimental results show good performance of the proposed architectures at low bit-rates.

We present a 3D modeling system for traffic surveillance applications. The application contains several cameras positioned around a traffic scene. The video signals are compressed and streamed to a central client. At the central client a scene reconstruction is carried out, using all 2D camera views utilizing 3D camera calibration information. To combine all views into a 3D model, moving objects are separated from static background first. Then a geometric model is built for the background. Foreground objects are modeled exploiting database objects. The appropriate textures are taken from the camera views and combined to form a multi-texture surface. Finally, lighting and view-dependent interpolation is carried out on the graphics card.