Capture and process 3D information with Kinect

Depth-sensing cameras offer cost-effective capabilities for developing 3D ComputerVision tools for specialized applications and vertical markets.

We develop ComputerVision software to use the Kinect for 3D scanning and measuring.

How does it work?

A depth camera measures the distance to each visible point on an object to create a collection of distances called a "depth Image" or "depth map." As a game controller, Kinect uses the depth image for real time gesture recognition.

We use the depth map information to compute X, Y and Z coordinates of each visible point on the object.

This collection of coordinates is called a "point cloud."

Typically, depth cameras create depth images at the rate of approximately 30 frames per second . In turn, our software generates around 30 point clouds per second.

When the camera and the object in its view are both stationary, the point cloud from one frame to the next will be the same. However, if either the camera or the object moves, a different set of points enters the visible area, the depth image changes, and we form a new point cloud.

A collection of point clouds has many uses, such as obtaining measurements and creation of 3D models. While you could measure distances using a single frame, we get more accurate results by applying mathematical algorithms on multiple frames to eliminate noise.

To create accurate 3D models using a depth camera, we generate multiple point clouds from multiple points of view. We then 'stitch' these point clouds together to form a single object. The last step is to remove the background information in order to arrive at an accurate, isolated object.

How can you use a 3D depth-sensing camera?

We'll help you design and develop the custom software needed to make your application work.