Actionscript 中使用相机视频进行手势识别
我想使用actionscript 3实现一个flex应用程序：使用相机和相机中的虚拟按钮（如图像），人们可以移动他的手越过虚拟按钮字段以触发按钮。现在我的解决方案是这样的：捕获erery帧中的按钮字段以获得bitmapdata，并计算bitmapdata的灰度总和，如果两帧之间的差异大于给定值，则触发 按钮。实际上，它可以工作，但它对背景的照明条件非常敏感，更糟糕的是，当有多个按钮时，人们经常会误触发按钮。还有更好的想法吗？有没有用于此目的的第三方图书馆？
I want to implement an flex application using actionscript 3: with the camera and the virture buttons(like a image) in the camera, one can move his hand to pass over the virture button field in order to fire the button. Now my solution is following this way: capture the button field in erery frame to get a bitmapdata, and calculate the gray-scale summation of the bitmapdata, if the difference between the two frames is greater than a given value, then fire
the button. Actually, it could work, but it is very sensible to the lighting condition of the background, worse, when there are more than one buttons, people often fire the button by misdike. Any better ideas ? Is there any 3rd-paty libary for this purpose ?
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（2）
您应该查看 http://www.bytearray.org/?p=3027
You should look at http://www.bytearray.org/?p=3027
我不确定这是否会很好地工作，因为您没有足够的有关相机的信息来进行准确的读数。它可能只有在完美的条件下才能工作。请记住，即使使用红外摄像头，Kinect 也需要一段时间才能使正确的硬件/软件组合正常工作。这只是一个白日梦。您需要某种独特的参考点。
I'm not positive this will work very well as you don't have enough information about the camera to have an accurate reading. It's probably only going to work in perfect condition. Remember that even with an infrared camera, the Kinect took a while to get the proper hardware/software combo to work. This is just a pipe dream. You need some sort of unique point of reference.