基于图像处理的运动测量方法与系统的研究/Study on Motion Measurement Method and System based on Image

2018-11-28 09:27:13

image measurement motion processing 图像处理



图像处理和计算机视觉在早期只局限于对静态图像的处理和分析,其实,动态景物在客观世界中更普遍,连续的图像序列能为我们提供更丰富的信息。动态图像分析的基本任务是从图像序列中检测出运动信息,识别与跟踪运动目标和估计三维运动及结构参数,它是图像处理与计算机视觉领域中的一个非常活跃的分支。
近年来,基于图像处理的检测系统已涉足到许多场所,比如重要场所的保安、汽车的自动驾驶和辅助驾驶、交通流量的控制、航空和军用飞行器的制导等。在工业生产中,图像处理的应用主要有产品质量检测,生产过程的自动控制,计算机辅助设计和计算机辅助制造等。
本文从计算机视觉应用的角度出发,将双目立体视觉应用于基于图像序列的运动测量系统中,并设计了相应的软件和硬件,计算空间三维刚体的运动信息,以实现主动视觉。作者无论是在测量系统的建模上,还是软件程序的编制上,以及测量系统的实验验证等,都做了许多积极的探索。
本文首先介绍了计算机图像处理的基本理论和序列图像的分析方法,然后重点论述了基于图像处理的运动测量系统的软件和硬件部分,主要包括以下几个方面:1、将双目立体视觉应用于运动测量系统中;2、提出了一种简单实用的特征点提取方法;3、对图像处理的有关算法,结合实际问题,进行了改进;4、将VB与Matlab相结合,开发了一套处理软件,既利用了Matlab的强大数值计算能力,又利用了VB的可视化友好界面,两者相得益彰。
在实验验证部分,首先对理想的摄像机进行了标定,然后通过对实际运动物体进行了测量实验,论证了该系统的基础理论完全正确,测量方案切实可行,为以后进一步的研究提供了良好的基础。



Image processing and computer vision were earlier limited exclusively to processing and analysis of static images. However, dynamic scenes are universally existent and serial image sequences could provide more comprehensive information. The essential purpose of dynamic image processing is to detect motion information from image sequences, to recognize and trace motion objects, and to estimate three-dimensional motion and structure parameters. This field has been a focus of interest in image processing and computer vision.
In recent years, image-processing-based detection systems have been used in various fields, such as security systems, automatic and auxiliary driving of automobiles, traffic control as well as guide of aerial and military aircrafts. In industry, image processing finds its important application in product quality control, production automation, computer- aided design and computer-aided manufacturing.
In this study, stereoscopic vision is applied to motion measurement system based on image sequences. Corresponding software and hardware have been developed to realize active vision by means of real-time tracing of the spatial motion of tridimensional rigid body. This study explores the modeling of motion measurement system, the development of related programs and the experimental verification of the measurement system.
The basic theory of image processing and the methods for analysis of image sequences are introduced at the beginning of the thesis. The description of software and hardware for the motion measurement system is divided into three parts: 1) The application of stereological vision to motion measurement system; 2) A convenient and practicable method was proposed for the determination of characteristic points; 3) Some improvement was made about related algorithms of image processing in realistic situation; 4) A processing software was developed by means of the combination of VB and Matlab, which made use of both the excellent computing capacity of Matlab and the friendly user interface of VB.
In the part of experimental verification, the validity of this motion measurement system is confirmed by the measurements on actual motion objects after the calibration of ideal camera. This measurement system is practical and provides a good foundation for future research.