Multi-view Dynamic Scene Modeling

نویسندگان

  • Li Guan
  • Marc Pollefeys
چکیده

Li Guan: Multi-view Dynamic Scene Modeling. (Under the direction of Marc Pollefeys.) Modeling dynamic scenes/events from multiple fixed-location vision sensors, such as video camcorders, infrared cameras, Time-of-Flight sensors etc, is of broad interest in computer vision, with many applications including 3D TV, virtual reality, medical surgery, markerless motion capture, video games, and security surveillance. However, most of the existing multi-view systems are set up in strictly controlled indoor environments, with fixed lighting conditions and simple background views. Many complications limit the technology in the natural outdoor environments. These include varying sunlight, shadows, reflections, background motion and visual occlusion etc. In this thesis, I address different approaches overcoming all of the aforementioned difficulties, so as to reduce human preparation and manipulation, and to make a robust outdoor system as automatic as possible. The main novel technical contributions of this thesis are as follows: a generic heterogeneous sensor fusion framework for robust 3D shape estimation; a way to automatically recover 3D shapes of static occluder from dynamic object silhouette cues, which explicitly models the static “visual occluding event” along the viewing rays; a system to model multiple dynamic objects shapes and track their identities simultaneously, which explicitly models the “inter-occluding event” between dynamic objects; and a scheme to recover an object’s dense 3D motion flow over time, without assuming any prior knowledge of the underlying structure of the dynamic object being modeled, which helps to enforce temporal consistency of natural motions and initializes more advanced shape learning and motion analysis. A unified automatic calibration algorithm for the heterogeneous network of conventional cameras/camcorders and new Time-of-Flight sensors is also proposed.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Multi-View Stereo Reconstruction with High Dynamic Range Texture

In traditional 3D model reconstruction, the texture information is captured in a certain dynamic range, which is usually insufficient for rendering under new environmental light. This paper proposes a novel approach for multi-view stereo (MVS) reconstruction of models with high dynamic range (HDR) texture. In the proposed approach, multi-view images are firstly taken with different exposure tim...

متن کامل

On Camera Calibration for Scene

We present a fully integrated active vision system for interpreting dynamic scenes. We argue that even if the ego-motion of the mobile vision system is known, a single view camera calibration cannot adequately support scene model acquisition and maintenance. It is shown that stable camera/grabber chain calibration can be achieved using a multi-view calibration process. With such calibration, a ...

متن کامل

Modeling and Visualization of Human Activities for Multicamera Networks

Multi-camera networks are becoming complex involving larger sensing areas in order to capture activities and behaviors that evolve over long spatial and temporal windows. This necessitates novel methods to process the information sensed by the network and visualize it for an end-user. In this paper, we describe a system for modeling and on-demand visualization of activities of groups of humans....

متن کامل

Image reconstruction of non-planar scenes from compressed multi-view measurements

We extend the domain of application of a method developed recently for joint reconstruction of l images, representing the same scene, from few multi-view measurements. While this method was initially designed for planar scenes, we show here that parametric smooth transformations can be used to handle more challenging scene structures. Our algorithm estimates one reference image common to all vi...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2010