Creating Stereoscopic (3D) Video from a 2D Monocular Video Stream
It is a challenge to generate stereoscopic (3D) video through a single moving camera under widely varying conditions. We propose an efficient approach to create true stereoscopic video from a monocular video stream captured under various moving conditions
- PDF / 677,815 Bytes
- 10 Pages / 430 x 660 pts Page_size
- 58 Downloads / 252 Views
Intelligent Automation, Inc, Rockville, MD 20855 2 Arizona State University, Tempe, AZ 85287
Abstract. It is a challenge to generate stereoscopic (3D) video through a single moving camera under widely varying conditions. We propose an efficient approach to create true stereoscopic video from a monocular video stream captured under various moving conditions. The approach contains three major steps. First, we apply Harris’ corner detector to detect distinctive feature points from a pair of image frames selected from the incoming video captured by a moving camera. Second, according to the consecutive property of the video, a local-window search based algorithm is developed for fast and accurate feature correspondence between the two image frames. Third, a hierarchical image rectification technique is designed to guarantee the success in creating a true and visually-comfortable stereo image for each incoming image frame. Besides, a software-based video stabilization algorithm is also developed for improved stereo video generation performance. Extensive tests using real video collected under various situations were performed for performance evaluation of the proposed approach.
1 Introduction In gaming and TV programs, 3D video effects are one of the most attractive features. 3D video techniques have also found wide civilian applications, such as medical operations, microscopy, scientific data display, and CAD/CAM. Military applications of 3D techniques include battlefield reconnaissance and surveillance. Conventional computer-based stereo vision techniques, although studied for many years, still have many limitations. For example, a conventional stereo vision system requires two identical cameras, a narrow baseline, fixed parameter settings and positions. It is only suitable for short-range scenes. However, in real world, camera motion is often nonstationary and viewpoints of the camera are different from time to time. Furthermore, parameter settings of the camera are sometimes unknown and variable during video capturing. To generate stereo (3D) video captured by a moving camera under widely varying conditions presents a challenge. The main reason is that the camera is obliquely mounted on a platform when the platform moves non-linearly or when the camera parameters vary while the platform is moving. In recent years, many research efforts have been made on stereo generation with uncalibrated cameras. Fusiello et al. [1] developed a compact and efficient stereo generation algorithm via image rectification. G. Bebis et al. (Eds.): ISVC 2007, Part I, LNCS 4841, pp. 258–267, 2007. © Springer-Verlag Berlin Heidelberg 2007
Creating Stereoscopic (3D) Video from a 2D Monocular Video Stream
259
However, their approach assumes that the stereo rig is calibrated, which means the intrinsic parameters of the camera pair such as focal length, aspect ratio, and their relative position are already precisely known. Unfortunately, as mentioned earlier, the camera parameters are not readily available and the relative position between
Data Loading...
