“物体放置在顶部”使用 OpenCV 进行检测
我是计算机视觉领域的新手，我想解决以下任务（最好使用 OpenCV 和 C#，但也非常欢迎使用 Scilab 等其他解决方案？）：
有一些参考对象，例如手（或多或少）静态）在场景中 - 相机俯视物体。现在我想识别我的手上是否有东西（它是否改变了我手的整体形状，或者它是否像坐在我的手掌上一样小）。
此任务仅用于演示目的，因此我想尽可能少地花费精力。我想用静态图片来训练它并在真实环境中使用它。
非常感谢任何解决此问题的帮助、提示或步骤。 先感谢您！
I'm new to the field of computer vision and I want to solve the following task (preferrably with OpenCV and C#, but other solutions like with Scilab? are also gratefully welcome):
There is some reference object like a hand (more or less static) in the scene - the camera is looking down on the object. Now I want to recognize if there is something on my hand (whether it changes the overall shape of my hand or if its just as small as sitting in my palm).
This task is for demonstration purposes only hence I want to use as less effort as possible. I'd like to train it with a static picture and use it in a real environment.
Any help, hint or steps how to tackle this problem are deeply appreciated.
Thank you in advance!
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。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
如果它基本上是静态的，那么我建议进行背景扣除。它将非常强大并且速度极快。
您可以运行高斯滤波器+阈值处理（固定阈值/大津型调整全局阈值/自适应局部阈值）来抓取差异图像中的斑点。这些斑点将表示变化，并且可能表示新的东西。
然后，您可以将其与原始检测到的手掌区域相交，以确定是否存在相交。
If it's largely static, then I'd recommend background subtraction. It will be very robust and blazingly fast.
You can run a Gaussian filter + thresholding (fixed threshold/Otsu-type adjusting global thresholds/adaptive local threshold) to grab blobs in the difference image. The blobs will denote change and likely something new.
You can then intersect that with the original detected palm region to figure out if there's an intersection.