OpenCV移动目标物体的检测跟踪的研究
本文对于实时的模拟码流图像采集是通过智能摄像机Kinect完成的，将其信息数据通过内部配置的高效AD转换装置转换为数字信息数据，然后将其传输给PC端，通过OpenCV视觉算法，进行帧图像的预处理，边缘检测，分割前景物和背景物，提取运动物体的轮廓，检测出运动物体的状态，并且进行精准的跟踪；并且最终在此基础上，引入了2D图像数据的3D重构，将其进行双目视觉的识别（由Kinect配备的红外摄像完成），并对其进行校正（运用Bouquet校正法），并且对于提取出的左右两幅图像中的运动物体进行检测，计算出上下、左右、前后的三文数据，并且结合OpenGL显示3D场景，最终实现检测3D空间场景下的运动物体的目的。
通过实验可以了解到，基于OpenCV算法的动态物体的检测、跟踪系统，能够精确的识别运动物体并且跟踪其运动。并且通过结合OpenGL的3D显示也可以达到描述三文空间的目的。
关键词：Kinect；双目视觉系统；三文重构；OpenCV；OpenNI；OpenGL；计算机视觉4046
Research of moving objects detection and tracking
ABSTRACT
This paper use intelligent camera Kinect to gain real-time acquisition simulation code flow image, by the internal configuration of efficient A\D conversion device will be the information
data is converted to a digital information data, the USB interface transfer to PC machine, through the OpenCV visual algorithm, carries on the collected frame image pretreatment, edge detection, segmentation foreground object and background objects, extracting motion objects outline, detect moving objects state, and accurate tracking; Finally, this paper based on this, the introduction of the 2-D image data of 3-D reconstruction, by Kinect equipped with infrared camera, binocular vision recognition, using Bouquet correction method carry on the correction, the correction of extraction after about two images of moving object detection, and calculates the up and down, left and right sides, before and after the three dimensional data, and combined with OpenGL show 3-D scene, so as to realize the in 3-D space scene of moving objects detection.
This paper summarizes the domestic and foreign development present situation of the intelligent computer vision technology , a deep research on the computer vision technology, image processing technology, dynamic monitoring technology, 3-D reconstruction technology and other related material. Detailed description of dynamic object detection system which is the based on OpenCV algorithm, computer vision technology and its design scheme, working principles and binocular vision technology to realize 3-D concrete realization process.
From the experimental results it can be seen that OpenCV algorithm based on dynamic object detection and tracking system can accurate recognition and tracking of moving object its movement. Combined with OpenGL 3-D display also completed the describe the function of three dimensional space.
Keywords: Kinect;Binocular vision system; Three dimensional reconstruction; OpenCV; OpenNI; OpenGL; computer vision
目 录
第1章 绪论 1
1.1 计算机视觉技术发展的历程 1
1.2 图像分析技术 1
1.3 国内外研究与应用现状 3
1.4 OpenCV发展趋势 5
1.5 论文中的研究内容 6
第2章 移动物体检测原理和关键技术 7