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
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.