海洋环境移动增强现实混合三维注册方法研究与应用

发布时间：2017-12-28 03:18

本文关键词：海洋环境移动增强现实混合三维注册方法研究与应用　出处：《上海海洋大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着海洋经济的发展,海洋灾害、海洋生态、海洋环境等问题给海洋科学的研究提出了许多具有挑战性的课题,这些问题需迫切得到解决,然而受我国海洋实情的制约,广大海洋科研人员缺乏充足的海洋数据,许多的研究工作无法顺利开展。因此,要打破这种僵局,我国应该大力加强和完善海洋监测能力建设,发展先进的海洋监测新技术来大量获取第一手丰富的海洋数据,从根本上加强我国海洋科技的研究基础。近海视频监测方式相对于传统监测方式具有成本低、适应性强、覆盖域广等优势,可及时了解近海动态变化的海况,但仅能显示肉眼能识别的基本海况,更为复杂的海洋信息需作后期处理;ROMS(Regional Ocean Model System)区域海洋模式作为一个新兴的海洋模式系统正被广泛应用于物理海洋、海洋地质、海洋声学等方面,其中物理海洋的应用可以预报海洋流场反映海洋流场基本要素,但其信息展示不如视频监测便捷、直观。因此,本文将移动视频监测方式与海洋流场预报信息相结合,利用当前流行的增强现实(Augmented Reality,简称AR)技术衍生出应用于海洋环境下的移动增强现实系统,直观、动态反映较为复杂的海况。增强现实系统性能的优劣直接取决于三维注册的精度,三维注册作用是检测用户所处的位置和视线方向,计算机由位姿信息计算虚拟信息在实际三维空间中的位置,虚拟信息实时叠加在视频序列上,从而增强视觉体验。主流的三维注册方式分为两种:基于计算机视觉和基于传感器,基于计算机视觉注册方式的匹配精度较高、不需要借助其他的硬件传感器,但是图像匹配算法较复杂,系统延迟性高、鲁棒性弱等;基于传感器的三维注册方式具有跟踪速度快、鲁棒性好等特点,但该三维注册方式受传感器精度影响较大,使得跟踪注册的精度不高。因此,单一三维注册方式的增强现实技术无法应对现实环境的复杂性,在注册精度和实效性上都受到限制,实际场景的应用需要综合各种跟踪定位方式来实现,这种综合多种跟踪定位的方式称为混合三维注册。为此,本文所做的主要工作为:1、针对单一三维注册方式很难解决海洋环境AR系统中跟踪注册的问题,提出了融合智能手机AGPS、电子罗盘和加速度传感器测定视线方向的多传感器注册与基于海天线特征标识的视觉跟踪注册的混合注册方法,以海天线标识辅助校正传感器测量的姿态角以提高三维注册精度。2、针对洋山港附近海域海面和天空对比度低的问题,本文提出了基于相位编组的Hough变换海天线提取算法来弥补单一Hough变换算法识别低对比度直线段能力弱的不足。3、以真实海洋环境及海洋数值预报为例,提出了适应海洋动态环境的增强现实框架,利用Android实现了原型系统,并对混合注册方法进行了验证。通过混合三维注册方式与单一传感器三维注册方式的精度对比分析以及系统运行时的数据传输效率测试,结果表明,此技术框架及混合三维注册方法有较强的可用性和实用性,在海洋动态环境的增强现实中具有广泛应用前景。
[Abstract]:Along with the development of marine economy, marine disasters, marine ecology, marine environmental issues and put forward many challenging topics for marine scientific research, these problems need to be solved urgently, but restricted by China's marine truth, the majority of marine scientific research personnel lack sufficient marine data, many research works can not be carried out smoothly. Therefore, in order to break this stalemate, China should strengthen and improve the capacity building of marine monitoring and develop advanced marine monitoring technologies to acquire first-hand abundant marine data and fundamentally enhance the research foundation of marine science and technology in China. Compared with the traditional video monitoring offshore monitoring method has the advantages of low cost, strong adaptability, wide coverage and other advantages, can understand the dynamic changes of the offshore sea, but only the basic sea eye can display recognition, more complex information required for marine processing; ROMS (Regional Ocean Model System) as a regional ocean model a new ocean model system is widely used in physical oceanography, marine geology, marine acoustics etc., which applied physics can forecast ocean ocean flow field reflect the basic elements of ocean flow field, but the video monitoring information display as easy and intuitive. Therefore, the mobile video monitoring mode and ocean flow forecast information combined with the current popular use of augmented reality (Augmented Reality, referred to as AR) technology applied to real mobile system derived from the marine environment is enhanced, intuitive, dynamic reflection of complex sea conditions. Enhance the pros and cons of the actual system performance depends directly on the accuracy of 3D registration, 3D registration function is to detect the user's position and view direction of computer by position information calculation in the actual position of the virtual information in 3D space, virtual real-time information superimposed on the video sequence, so as to enhance visual experience. The main way of 3D registration is divided into two types: Based on computer vision and sensor based on computer vision registration matching accuracy, without the help of other hardware sensors based on image matching, but the algorithm is complex, the system delay high, weak robustness; 3D registration method based on sensor has the characteristics of fast tracking and good robustness, but the registration accuracy is greatly influenced by the sensor, the tracking accuracy of registration is not high. Therefore, the complexity of augmented reality technology in single mode 3D registration can not cope with the reality, is restricted in the registration accuracy and effectiveness of the application, the actual scene needs a comprehensive range to achieve tracking and positioning, the comprehensive tracking and positioning is called hybrid 3D registration. Therefore, the main work of this paper is: 1, for a single 3D registration is difficult to solve the registration problem in the marine environment AR system, put forward the multi-sensor registration AGPS, intelligent mobile phone gaze direction determination of electronic compass and accelerometer and hybrid registration method of sea antenna characteristic identification based on visual tracking. The sea antenna marking auxiliary correction sensor attitude angle to improve the 3D registration accuracy. 2, in view of the low sea surface and sky contrast near the Yangshan Port, this paper proposes a Hough transform sea antenna extraction algorithm based on phase marshalling to compensate for the weakness of single Hough transform algorithm in identifying low contrast straight lines. 3, taking the real marine environment and marine numerical prediction as an example, we propose an augmented reality framework to adapt to the marine dynamic environment. The prototype system is implemented by Android, and the hybrid registration method is verified. By mixing the 3D registration methods of data transmission efficiency test, operation accuracy comparison and the analysis of the system with a single sensor of the 3D registration results show that the framework and hybrid 3D registration method has the usability and practicability, in enhancing the marine dynamic environment has wide application prospect in reality.
【学位授予单位】：上海海洋大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P714;TP391.41

【相似文献】