面向增强现实浏览器的全景图精确标注

发布时间：2018-02-03 06:38

本文关键词： 移动增强现实全景图精确标注用户体验　出处：《北京理工大学》2015年硕士论文　论文类型：学位论文

【摘要】：近年来，随着科技的飞速发展，高性能的移动终端为增强现实在移动领域的实现提供了有利条件。然而，现有的移动增强现实浏览器不够完善，，在一些方面仍有待改进。例如，目前大多数浏览器都是基于GPS和传感器数据进行POI的标注，标注方式精度较低；局限于移动设备的视场角，用户需手持设备，旋转以获取不同方向的POI数据，用户体验欠佳；现有的交互方式较为单一，以签到、分享地点为主，用户参与度不高。针对目前移动增强现实浏览器的不足之处，本文就以下方面开展了研究：（1）针对增强现实标签叠加不准确的问题，提出一种基于全景图的POI标注方式，利用传感器数据与模板匹配算法对POI进行多重定位。首先，建立全景图像素坐标系与世界地理坐标系的映射关系，以GPS和传感器数据计算当前地点与POI的方位角，将POI映射至全景图中。在实现了POI粗定位的基础上，以模板匹配算法进一步定位POI在全景图中的最佳匹配位置，实现像素级精确标注。同时，提出一种基于方位角减少模板匹配范围的方法，进一步提高POI标注的效率。实验证明，结合了传感器数据与模板匹配算法的POI标注方法，能有效提高POI的标注精度，改善标签叠加不准确的现象。（2）提出两种方式改进用户体验。首先，针对移动设备视场角窄小、用户需长时间手持设备的问题，提出一种基于全景图的POI离线浏览模式。通过左右滑动全景浏览已标注好的周围所有的POI，克服视场角窄小的缺陷。第二，针对交互方式单一的问题，设计了一种由用户补充添加自定义POI信息的交互模式。前一个用户可在圈出自自己感兴趣的区域作为自定义POI模板，并上传至云端服务器，作为后续使用者的模板图像，从而实现自定义标注信息的多人共享。在实际使用中，以上两种方式均能有效改善用户对增强现实浏览器的使用感，提高用户参与度。（3）基于以上研究工作，设计了一种基于全景图的增强现实的精确标注浏览器，在Android平台实现全景图拼接、POI标注、用户自定义添加POI、网络交互等功能。在实现基本功能的同时，针对移动端特性对图像拼接与模板匹配算法进行了优化。用户测试结果表明，本系统达到预期设计目标，实现了一个用户体验良好的移动增强现实浏览器。
[Abstract]:In recent years, with the rapid development of science and technology, high-performance mobile terminals provide a favorable condition for the realization of augmented reality in the mobile field. However, the existing mobile augmented reality browsers are not perfect. For example, most browsers are based on GPS and sensor data to annotate POI. Limited to the field of view of mobile devices, users need handheld devices to rotate to obtain different directions of POI data, the user experience is poor; The existing interaction mode is relatively single, mainly to sign in, share the place, the user participation is not high. In view of the shortcomings of the current mobile augmented reality browser, this paper has carried out research on the following aspects: 1) aiming at the problem of inaccurate label superposition of augmented reality, a panorama based POI tagging method is proposed, which uses sensor data and template matching algorithm to locate POI multiple times. The mapping relationship between the panoramic pixel coordinate system and the world geographical coordinate system is established, and the azimuth between the current location and the POI is calculated by using the GPS and sensor data. The POI is mapped to the panoramic image. On the basis of realizing the POI rough location, the template matching algorithm is used to locate the best matching position of POI in the panoramic image, and the pixel level accurate annotation is realized at the same time. A method based on azimuth to reduce the range of template matching is proposed to further improve the efficiency of POI tagging. Experimental results show that the POI tagging method combines sensor data with template matching algorithm. It can effectively improve the labeling accuracy of POI and the phenomenon of label superposition inaccuracy. Firstly, aiming at the narrow angle of view field of mobile device, the user needs a long time handheld device. This paper presents a POI offline browsing mode based on panoramic images. By using left and right sliding panorama to browse all the labeled POIs, the defect of narrow angle of view is overcome. Secondly, the problem of single interaction mode is addressed. An interactive mode of adding custom POI information to users is designed. The former user can be used as a custom POI template in the area of interest and upload to the cloud server. As a follow-up user of the template image, so as to achieve the customized annotation information sharing. In actual use, the above two methods can effectively improve the user's sense of use of augmented reality browser. Increase user participation. 3) based on the above research work, a panoramic augmented reality (AR) accurate annotation browser is designed, and the panoramic image splicing and POI tagging is realized on Android platform. User custom add POI, network interaction and other functions. In order to achieve the basic functions, the image stitching and template matching algorithm is optimized according to the characteristics of the mobile side. The user test results show that. The system achieves the desired design goal and realizes a mobile augmented reality browser with good user experience.
【学位授予单位】：北京理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;TP393.092

【参考文献】