Dissertation > Excellent graduate degree dissertation topics show

Study of Dense Depth Acquisition Techniques Based on Structured Light and Stereo Matching

Author: XuZhiLiang
Tutor: MaLiZhuang
School: Shanghai Jiaotong University
Course: Computer Science and Engineering
Keywords: Binocular stereo matching Image depth acquisition 3D model reconstruction Dynamic Programming Vertical restraints Structured light Pseudo-random sequence
CLC: TP391.41
Type: Master's thesis
Year: 2009
Downloads: 170
Quote: 5
Read: Download Dissertation

Abstract

The image depth access to technology goal is to recover from one or more images of three-dimensional objects in the scene geometry information. In recent years, the technology has been widely used in the field of industrial automation, virtual reality, computer-aided design (CAD) and digital entertainment. The stereo matching is a classical depth Get it by matching the captured multiple images of the same scene at different positions in pixels or the feature point corresponding relationship obtained with respect to the distance from the camera, and thus to recover from the two-dimensional image out three-dimensional scene depth information. Because the technology is practical, efficient, high degree of automation, it is attracting much attention in the field of computer vision and computer graphics research focus. Access to technology can be divided into two directions - active stereo matching and passive stereo matching based on the depth of the stereo matching. Active stereo matching method is input to the scene lighting information, for example, a laser scanning line, structured light, increasing the identifiable characteristic scene, thereby reducing the difficulty of matching; passive stereo matching scenes do not need to add any auxiliary information . In this paper, the advantages of active stereo matching and passive stereo matching, the proposed acquisition method based on two-dimensional pseudo-random structured light pattern and dense stereo matching images dense depth. It requires only one picture scene, the depth information of each pixel in the image can be obtained, and can handle the lack of texture in the scene. The method of process is divided into three steps. First, generate a two-dimensional binary pseudo-random structured light pattern map, and use common projector projection to the scene of the mode plan. Then, the structure on the camera light scene image radial distortion correction and stereo matching algorithm using dynamic programming matching information of the corresponding pixel. Finally, the geometric parameters of the camera and projector calibration obtained using the least squares method based on the triangulation method to obtain dense depth of the scene. At the same time, we stereo matching based on the conventional dynamic programming algorithms, a longitudinal constraint is added in the calculation of the optimal matching path constraints, i.e. by matching path to the previously calculated current optimal matching path, thereby eliminating the scanning line in the depth map phenomenon, and stereoscopic vision on the site of the Middlebury College listed in accuracy over most of binocular stereo matching algorithm with the same computational efficiency. The only use projectors, cameras and other simple devices to obtain dense depth information of the scene, less human intervention, automatically. The experimental results show that this method can get high quality dense depth, simple laboratory equipment to achieve reasonable accuracy for a wide range.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net Mobile