无需标记的增强现实应用

发布于 2024-11-14 10:01:39 字数 1436 浏览 5 评论 0原文

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。

评论(3

凯凯我们等你回来 2024-11-21 10:01:39

现在有几个可用的自然特征跟踪库:Qualcomm AR SDK、Metaio Unifeye SDK 和 Metaio Junaio Glue 都允许您在移动设备上注册和跟踪图像,HIT Lab NZ 拥有提供类似功能的 Opira 库但在桌面上(具有非常好的创作工具)。

将 3D 对象插入到未准备、未建模的视频流中的主要方法有两种:第一种是对 SLAM(同时定位和建图)进行一些变体 - 查找视频流中的特征点,检测帧到帧之间稳健的特征点。帧,并使用它们构建点的 3D 地图,该地图可用作插入 3D 对象的环境/跟踪模型。 AR 圈中最著名的是 PTAM(并行跟踪和地图)。第二种方法主要用于户外:假设用户保持静止,并要求他们生成全景图,将其投影到用户周围的圆柱体上。然后可以将对象插入到圆柱体周围的坐标中,并且可以学习和跟踪环境。这是全景跟踪和地图(令人困惑的是,相同的缩写)。

希望这可以帮助您入门!

There are a couple of natural feature tracking libraries becoming available now: the Qualcomm AR SDK and Metaio Unifeye SDK and Metaio Junaio Glue all allow you to register and track an image on mobile devices, and HIT Lab NZ has the Opira library which provides similar functionality but on desktop (with really nice authoring tools).

There are two main approaches to inserting a 3D object into an unprepared, unmodelled videostream: the first is to do some variation on SLAM (simultaneous location and mapping) - find feature points in the videostream, detect the ones that are robust frame-to-frame, and use those to build a 3D map of points, which can be used as an environment/tracking model for inserting your 3D object. The most famous in AR circles is PTAM (parallel tracking and mapping). The second approach is mostly used outdoors: assume the user will stay still, and ask them to generated a panorama, which you project onto a cylinder around the user. Then objects can be inserted into coordinates around the cylinder, and the environment can be learned and tracked. This is Panoramic Tracking and Mapping (confusingly, the same acronym).

Hope this helps you get started!

瑶笙 2024-11-21 10:01:39

您可以尝试Sony SmartAR SDK来实现无标记的AR,包括高速摄像头和大规模。不幸的是,该项目看起来被放弃了。该文档可以在下载的 zip 文件中找到。

https://www.sonydna.com/sdna/solution/SmartAR_SDK.html

这是一个令人难以置信的演示:https://www.youtube.com/watch?v=XCEp7udJ2n4

更好的是,有一个更新的无标记 AR SDK,名为 Kudan:https://www.kudan.eu/features/

您可以选择使用 Unity 3D 软件包。

You can try Sony SmartAR SDK witch implement AR without markers, including high speed camera and large scale. Unfortunately the project looks abandoned. The documentation will be found in the downloaded zip file.

https://www.sonydna.com/sdna/solution/SmartAR_SDK.html

And here is an incredible demo: https://www.youtube.com/watch?v=XCEp7udJ2n4

Even better, there is an updated markless AR SDK called Kudan: https://www.kudan.eu/features/

You have the option to use the Unity 3D package.

〃温暖了心ぐ 2024-11-21 10:01:39

这是一个正在进行的研究领域。据我所知,目前还没有已知的稳健(且快速)的算法。也许您可以在简历论文上找到一些灵感。

This is a field of ongoing research. As far as I know, there are no known robust (and fast) algorithms yet. Maybe you can find some inspiration on CV Papers.

~没有更多了~
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文