Depth stream motion tracking: 5 things to watch out for

Predators have eyes at the front to make it easier to discern prey.

Why did an article about depth stream motion tracking start with this strange quote? Because the advantage of having eyes at the front of our heads is that we can discern depth. This means we can see in 3D. Animals that have eyes on the sides of their heads (a chicken for example), cannot discern depth as precise as, say, a leopard can. On the other hand, their angle of view is much wider than leopard’s (or ours as a matter of fact).

Depth cameras, with their depth stream, give us a much richer information about their surroundings than video cameras.  They do have some disadvantages though. Some of these will be discussed in this article.

Depth cameras have gained popularity especially since the introduction of Microsoft Kinect in 2010. It is a product allowing users to interact with technology with gestures. Nowadays there are other products on the market as well, such as Asus Xtion or Leap motion controller to name a few.

How do depth cameras work?

Depth cameras use a variety of approaches to obtain a depth image of the space. Now, what is a depth image? It’s basically a grayscale image where the color of a pixel corresponds to the distance of that pixel from the camera.

Original kinect depth stream on the left, extracted silhouette on the right.

Most consumer-grade depth cameras rely on the time-of-flight approach, which measures the time it takes for the light to come out of the camera, hit the object and bounce back. For those of you who want to delve into other approaches, they are called structured light, triangulation and modulated light.

What to be mindful of at depth stream motion tracking

Range

Most depth cameras have a limited range of around 4m. Care has to be taken when determining the exact location of the depth camera in relation to the area that you want to track.

Protection against environment

Depth cameras by and large don’t have any weather protection because they have been designed primarily for indoor operation.

Light conditions

Depending on the type of depth camera, there may be some limitations of the light conditions it can be used in. For example, the performance of Microsoft Kinect drops significantly in direct sunlight, because the IR sensor gets flooded by surrounding IR light.

USB connection

Most depth cameras are connected to a computer with a USB connection. As they need a very large bandwidth to transfer all the data, the range of a standard USB cable can only be a couple of meters. If you need to install your depth camera further away from the controller you should use a USB extender.

Framerate

Tracking framerate means how many depth images per second the tracking algorithm analyses. In addition, whenever we are using a depth camera, the algorithm usually works on 2 sets of streams, video stream as well as depth stream, which means framerate has a big effect on the required processing power of the controller.


This is an excerpt from our guidebook on Technologies for Interactive Architecture. Check it out!

Related Posts
Let’s talk!