運用移動向量之影像深度偵測演算法

Depth Detection Algorithm using Motion Vector

古全鈞(Chuan-Chun Ku)

指導教授 ： 龔志賢

長榮大學/資訊暨工程學院/資訊管理學系(所)/碩士(2012年)

https://doi.org/10.6833/CJCU.2012.00205

摘要

立體視覺一直是電腦視覺領域的重要研究領域之一，近幾年來，隨著科技快速發展進步，立體視覺技術已經被廣泛的應用在各個領域，例如:即時影像追蹤、機器人導航、虛擬實境等。本篇論文將以電腦為平台，配置雙攝影機擷取即時影像，搭配本論文提出之運用移動向量之影像深度偵測演算法計算影像深度，建構出即時立體影像深度計算系統。
由於攝影機擷取之影像，格式為MPEG1、MPEG2與MPEG4等，皆使用移動預估與移動補償技術（Motion Estimation/Motion Compensation），因此每一幀（Frame）中皆有移動向量（Motion Vector）資訊。本論文提出之方法，首先針對第一張影像做計算，取得所需移動向量資訊後，並與第二張影像之移動向量做比對，進而計算出影像深度。所提出之運用移動向量之影像深度偵測演算法只需計算移動向量後進行比對計算，因而可以快速有效率地取得影像深度，可供未來更快速三維影像重建、2D轉換3D影像或機器人導航等使用。

關鍵字

立體視覺 ； 三維重建 ； 機器人導航

並列摘要

In this thesis, the design and the construction of an innovative system for stereo vision using the motion vector are described. Images captured by the camera, the format of MPEG1, MPEG2 and MPEG4, are using motion estimation and motion compensation technologies. Two superimposed stereo images are received simultaneously as a complex image and using the propose algorithm, the disparities and depth of objects of the scene can be obtained. Since taking images of a single object from various angles by two cameras and finding the same points in pictures through the proposed algorithm, the same point has different position in different pictures. This thesis proposes an innovative scheme using the motion vector to restore the actual coordinates of positions. Based on the data of the previous and current images, the proposed scheme could take less time to calculate the depth of the image effectively. In the future, the proposed scheme can be employed for the three-dimensional image reconstruction, 2D to 3D image conversion and robot navigation.

並列關鍵字

stereo vision ； three-dimensional image reconstruction ； robot navigation

參考文獻

[2] D. Marr and T. Poggio, “A Theory of Human Stereo Vision,” A. I. Memo No. 451, November 1977.

[5] Nikolaos P. Papanikolopoulos, “Visual Tracking of a Moving Target by a Camera Mounted on a Robot: A Combination of Control and Vision,” IEEE Transactions on Robotics and Automation feabruary 1993.

[6] Brett Browning and Manuela M. Veloso, “Real-Time, Adaptive Color-based Robot Vision,” Carnegie Mellon University, August 2005.

[7] Ezio Malis and Eric Marchand, “Experiments with robust estimation techniques in real-time robot vision,” IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, IROS'06, Beijing, China, October 2006.

[9] Theodore Pachidis and John Lygouras, “A Pseudo Stereo Vision System as a Sensor for Real Time Path Control of a Robot,” IEEE Instrumentation and Measurment Technology Conference, Anchorage, AK, USA, 21-23 May 2002.