https://blog.csdn.net/weixin_46450859zh-cnhttps://blog.csdn.net/https://blog.csdn.net/weixin_46450859/article/details/149916407https://blog.csdn.net/weixin_46450859/article/details/149916407weixin_46450859Mon, 04 Aug 2025 23:20:34 +0800https://blog.csdn.net/weixin_46450859/article/details/149333354https://blog.csdn.net/weixin_46450859/article/details/149333354weixin_46450859Mon, 14 Jul 2025 16:19:33 +0800https://blog.csdn.net/weixin_46450859/article/details/148614752https://blog.csdn.net/weixin_46450859/article/details/148614752weixin_46450859Thu, 12 Jun 2025 17:26:51 +0800https://blog.csdn.net/weixin_46450859/article/details/148436773https://blog.csdn.net/weixin_46450859/article/details/148436773weixin_46450859Wed, 04 Jun 2025 23:24:06 +0800https://blog.csdn.net/weixin_46450859/article/details/145771087https://blog.csdn.net/weixin_46450859/article/details/145771087weixin_46450859Fri, 21 Feb 2025 10:28:09 +0800https://blog.csdn.net/weixin_46450859/article/details/138508302https://blog.csdn.net/weixin_46450859/article/details/138508302weixin_46450859Mon, 06 May 2024 21:20:51 +0800https://blog.csdn.net/weixin_46450859/article/details/138243735https://blog.csdn.net/weixin_46450859/article/details/138243735weixin_46450859Sat, 27 Apr 2024 10:52:14 +0800100，且和mInitialFrame成功匹配的特征点数量>=100，那么将会调用初始化器进行两帧估计位姿，以及生成3D点，关键的来了，这样就可以解决单目初始化的连续两帧视差过小的问题。]]>https://blog.csdn.net/weixin_46450859/article/details/137053714https://blog.csdn.net/weixin_46450859/article/details/137053714weixin_46450859Tue, 26 Mar 2024 20:12:29 +0800https://blog.csdn.net/weixin_46450859/article/details/136951546https://blog.csdn.net/weixin_46450859/article/details/136951546weixin_46450859Fri, 22 Mar 2024 21:22:54 +0800https://blog.csdn.net/weixin_46450859/article/details/134011752https://blog.csdn.net/weixin_46450859/article/details/134011752weixin_46450859Tue, 24 Oct 2023 15:00:10 +0800setFixed(false);其中T1为第一张图片的相机位姿，T2为第二张相机位姿，pts_1为目标点在第一张图片的归一化平面位置，就是把我们得到的像素坐标转到归一化平面，pts_2为目标点在第2张图片，pts_4d就是我们需要得到的3D位置，不过需要把最后一维归一后，取前三维数。好了，现在我们有了相机外参T，相机内参已知，目标点（mark中心）的像素坐标。我们就可以用opencv的三角化求目标点的3D位置了，]]>https://blog.csdn.net/weixin_46450859/article/details/133906358https://blog.csdn.net/weixin_46450859/article/details/133906358weixin_46450859Wed, 18 Oct 2023 15:15:30 +0800https://blog.csdn.net/weixin_46450859/article/details/133799263https://blog.csdn.net/weixin_46450859/article/details/133799263weixin_46450859Thu, 12 Oct 2023 21:28:48 +0800https://blog.csdn.net/weixin_46450859/article/details/132683502https://blog.csdn.net/weixin_46450859/article/details/132683502weixin_46450859Tue, 05 Sep 2023 14:04:47 +0800https://blog.csdn.net/weixin_46450859/article/details/132680172https://blog.csdn.net/weixin_46450859/article/details/132680172weixin_46450859Tue, 05 Sep 2023 00:27:29 +0800https://blog.csdn.net/weixin_46450859/article/details/132678352https://blog.csdn.net/weixin_46450859/article/details/132678352weixin_46450859Mon, 04 Sep 2023 22:06:30 +0800