摘要计算机双目立体视觉是通过多幅二维的平面图像恢复出被拍摄目标物体的三维 信息,其基本原理是用摄像机来模仿人眼与人类视觉的立体感知过程,从两个不同的 视点来拍摄现实场景中的同一目标物体,获取不同视角下的感知图像,并通过三角测 量的方法来计算图像像素间的位置偏差,以获取景物的三维深度信息,该技术在许多 领域中都具有广泛的应用。86023
通常来说,一个较为完整的双目视觉系统是由图像获取、摄像机标定、角点检测、 立体匹配、目标定位和深度图生成六大部分组成的。而我的毕业设计课题是对摄像机 标定内容展开研究并对图像进行加工与处理。
本次毕业设计是团队项目基于双目测距系统前端设计与开发,是在 VS2010 及 opencv2。4。9 的基础上进行开发的,然后将在自己电脑上成功运行的程序移植到高性 能嵌入式开发板 Jetson-tk1 中,使得最终成品能够完成毕业设计功能要求。
Jetson-tk1 开发者套件针对嵌入式系统应用释放 GPU 的潜能。它以性的 NVIDIA Tegra® K1 SoC 为基础构建,并且使用相同的 NVIDIA Kepler™计算核心, 用于快速开发和部署面向计算机视觉、机器人技术、医疗和更多领域的计算密集型系 统。
毕业论文关键词:双目测距;Jetson-tk1;OpenCv;
Abstract Computer binocular stereo vision is the great picture of two dimensional plane image restoration was shooting target 3 d information of the object, its basic principle is to use the camera to mimic the human eye and human visual stereo perception process, from two different viewpoints to the same target objects in real scenes, get the perception of the image under different Angle of view, and through the triangulation method to calculate the position deviation between image pixels, to get a scene of 3 d depth information, the technology has wide application in many fields。
Generally speaking, a relatively complete binocular vision system consists of image acquisition, camera calibration, the corner detection, stereo matching, target positioning and depth map generation composed of six parts。 My graduation design topic is the study of content of camera calibration and image processing and processing。
This graduation design is a team project front-end design and development, based on the binocular distance measurement system is conducted on the basis of in VS2010 and opencv2。4。9 development, then your program is running on your computer successfully transplanted into the high-performance embedded development board Jetson - tk1, enables the finished products to complete the graduation design function requirement。
Jetson - tk1 developer kits for embedded system application to release the potential of the GPU。 Its revolutionary NVIDIA prototypes ® K1 SoC based building, and use the same NVIDIA Kepler ™ to calculate core, used for rapid development and deployment for computer vision, robotics, medical and more computationally intensive in the field of system。
Keywords: Binocular Ranging;Jetson-tk1;OpenCv;
目 录
第一章 绪论 1
1。1 课题背景 1
1。3 选题的目的和意义 2
1。4 本文研究的主要内容 3
第二章 平台简介及需求分析 4
2。1 OpenCv