Shape-from-texture revisité : reconstruction 3D et auto-calibrage à partir d’une seule image Shape-from-Texture Revisited : 3D-reconstruction and Self-calibration from a Single Image
نویسندگان
چکیده
Dans cet article, nous revisitons le problème du shapefrom-texture. Notre première contribution est d’ordre méthodologique. Nous montrons que l’estimation de la profondeur et de la normale est équivalente à un calcul de pose. Notre deuxième contribution consiste à utiliser la redondance d’information entre la profondeur et la normale pour auto-calibrer la caméra. Nous validons ces deux contributions sur images de synthèse et sur image réelle.
منابع مشابه
Structured light on dynamic scenes using standard stereoscopy algorithms
Most 3D reconstruction systems based on structured light use a single camera for sensing, and a projector that emits one or several patterns over the observed surface. We use the following analogy between these systems and a classical two-camera stereoscopic system: the projector can be seen as one camera for which the image of the 3D scene is known (it is the emitted pattern), and the other ca...
متن کامل3D reconstruction of real objects with high resolution shape and texture
We present a robust and accurate system for 3D reconstruction of real objects with high resolution shape and texture. Our reconstruction method is passive, the only information needed being 2D images obtained with a calibrated camera from different view angles as the object rotates on a turntable. The triangle surface model is obtained by a scheme combining octree construction and marching cube...
متن کاملDeep Shape from Silhouettes
Despite strong progress in the field of 3D reconstruction from multiple views, challenges from noise, transparencies and lack of texture remain. On the other hand, silhouette based reconstruction techniques ease the dependency of 3d reconstruction on image observations but need a large number of silhouettes as well as calibration information. In this paper, a novel end to end pipeline is propos...
متن کامل3D Face Reconstruction from Single Two-Tone and Color Images
This chapter addresses the problem of reconstructing the three-dimensional shape of faces from single images. We present an algorithm that uses prior knowledge of faces—a single shape model of a face—to eliminate the ambiguities in the reconstruction [13, 14]. The algorithm achieves veridical reconstruction results on images taken under a wide range of viewing conditions. In addition, it can re...
متن کاملNear Light Correction for Image Relighting and 3D Shape Recovery
In this paper, we propose a near-light illumination model for image relighting and 3D shape recovery. Classic methods such as used by the popular RTI software from Cultural Heritage Imaging assume that lighting is infinitely far away from the scene. However, this constraint is impossible to achieve in practice: light sources cannot be too far away from the scene due to space and power constrain...
متن کامل